refactor: update module path to git.pengzhan.dev/haopengzhan/hireme across all internal packages and imports

.gitignore

@@ -1,3 +1,4 @@
 data/db
 vendor/
 build/
+.env

.vibe/ai-guideline.md

@@ -0,0 +1,32 @@
+# AI Development Guidelines for Hireme
+
+These guidelines provide context to AI assistants working on the `hireme` repository. Adhere to these principles to maintain the project's architecture and design philosophy.
+
+## 1. Single Binary & Zero Dependencies
+
+- **Goal:** The application must compile into a single Go binary.
+- **Rule:** Do NOT introduce external dependencies that require separate build steps, node_modules, or static asset folders unless explicitly authorized.
+- **Rule:** Use Go 1.22+ standard library features wherever possible (e.g., `net/http` `ServeMux`).
+
+## 2. UI Philosophy (HTMX + Tailwind)
+
+- **Rule:** Do NOT introduce React, Vue, or any heavy JS frameworks.
+- **Rule:** All dynamic interactions must be handled via HTMX (`hx-get`, `hx-post`, `hx-target`). Endpoints should return HTML fragments, not JSON, for UI updates.
+- **Rule:** CSS is handled via Tailwind CSS utility classes. Maintain the dark mode aesthetic using `bg-slate-*`, `text-slate-*`, and vibrant accents (`indigo`, `emerald`, `rose`).
+- **Rule:** UI templates are stored as constant strings in Go files (`internal/ui/template_*.go`) and registered in `templates.go`. Do not create external `.html` files.
+
+## 3. Database Integrity
+
+- **Rule:** Use `go-sqlite3` and ensure `PRAGMA journal_mode=WAL;` is maintained. The application heavily relies on WAL mode to allow concurrent reads (from the UI) and writes (from background workers).
+- **Rule:** Avoid long-running transactions that could lock the database.
+
+## 4. Structured Logging
+
+- **Rule:** Use `zap` structured logging exclusively. Do NOT use the standard library `log.Printf`.
+- **Rule:** Use the `w` methods (`Infow`, `Debugw`, `Errorw`) to pass key-value pairs.
+  - **Correct:** `logger.Log.Infow("User created", "component", "API", "user_id", id)`
+  - **Incorrect:** `logger.Log.Infof("[API] User created: %d", id)`
+
+## 5. Background Workers
+
+- **Rule:** Long-running tasks (Crawling, LLM Analysis) must be executed asynchronously via the `internal/worker` pool. Do not block HTTP request handlers. Enqueue tasks in the SQLite `tasks` table instead.

.vibe/skills/logging.md

@@ -0,0 +1,51 @@
+---
+name: Structured Logging with Zap
+description: How to correctly use the zap structured logging library in this repository instead of standard log or string formatting.
+---
+
+# Structured Logging with Zap
+
+When adding logs to the `hireme` repository, **always** use the global `logger.Log` (which is a `zap.SugaredLogger`).
+
+Do **NOT** use `log.Printf` or `fmt.Printf`.
+Do **NOT** use `logger.Log.Infof` to format strings with variables if those variables should be structured data.
+
+## Proper Usage (The `w` methods)
+
+Use the `w` variants of the log methods (`Infow`, `Debugw`, `Errorw`, `Warnw`). These methods take a message string as the first argument, followed by variadic key-value pairs.
+
+### Example: API Endpoints
+
+Always include a `"component", "API"` field.
+
+```go
+// GOOD
+logger.Log.Infow("Triggered endpoint", "component", "API", "endpoint", "/applications", "status", "success")
+
+// BAD - String interpolation hides the data from JSON parsers
+logger.Log.Infof("[API] Triggered /applications with status %s", "success")
+```
+
+### Example: Background Workers
+
+Include the worker ID and task ID.
+
+```go
+// GOOD
+logger.Log.Errorw("Task processing failed", "component", "Worker", "worker_id", id, "task_id", task.ID, "error", err)
+```
+
+### Example: Database Operations
+
+```go
+// GOOD
+logger.Log.Debugw("Running query", "component", "DB", "query", "SELECT * FROM jobs")
+```
+
+## Summary of `w` Methods
+
+- `logger.Log.Debugw(msg string, keysAndValues ...interface{})`
+- `logger.Log.Infow(msg string, keysAndValues ...interface{})`
+- `logger.Log.Warnw(msg string, keysAndValues ...interface{})`
+- `logger.Log.Errorw(msg string, keysAndValues ...interface{})`
+- `logger.Log.Fatalw(msg string, keysAndValues ...interface{})` // Note: calls os.Exit(1)

Makefile

@@ -2,7 +2,7 @@
 
 BINARY_NAME=hireme
 CONFIG_PATH=data/config/targets.yaml
-PORT=8080
+PORT=12345
 ENV=dev
 DB_PATH=data/db/$(ENV)/hireme.db
 

README.md

@@ -0,0 +1,66 @@
+# Hireme (GJCC)
+
+**Hireme (Google Job Command Center)** is a high-performance, single-binary job application and tracking command center written in Go.
+
+It completely automates the job search workflow by crawling job boards, analyzing job descriptions against your profile using Large Language Models (LLMs), and presenting a sleek, HTMX-powered UI to review, save, and track your applications.
+
+## Key Features
+
+- **Automated Crawling:** Background worker pools discover and fetch job listings from targets (e.g., Google Jobs).
+- **LLM Job Analysis:** Integrates with LLMs (Mock, Gemini, Claude) to score jobs based on your qualifications, identify missing skills, and provide resume tailoring suggestions.
+- **Smart Filtering:** Multi-stage filtering pipeline (Hard Keywords + Scoring Weights) configured via YAML to automatically reject unqualified jobs before expensive LLM calls.
+- **Sleek UI:** A responsive, dark-mode dashboard built with Go Templates, HTMX, and Tailwind CSS.
+- **Pipeline Management:** An interactive Apply Table to move jobs through your pipeline (Saved ▶ Applied ▶ Interview ▶ Offer).
+- **Single Binary & SQLite:** Everything is packaged into a single Go binary using a local SQLite database (WAL mode enabled) for zero-dependency deployment.
+
+## Quick Start
+
+### Prerequisites
+
+- Go 1.26+
+- `make`
+
+### Installation
+
+1. **Clone the repository:**
+
+   ```bash
+   git clone https://git.pengzhan.dev/haopengzhan/hireme.git
+   cd hireme
+   ```
+
+2. **Configure Data & Filters:**
+   Copy the example configurations and adjust them:
+   - `data/config/targets.yaml`: Setup crawler targets.
+   - `data/config/filters.yaml`: Setup hard keywords and scoring weights.
+
+3. **Set Environment Variables:**
+
+   ```bash
+   export LOG_LEVEL=debug
+   export GEMINI_API_KEY="your_api_key" # If using Gemini LLM
+   ```
+
+4. **Build and Run:**
+
+   ```bash
+   make clean
+   make run
+   ```
+
+5. **Access the UI:**
+   Open your browser and navigate to [http://localhost:8080](http://localhost:8080).
+
+## Commands
+
+- `make build`: Builds the `hireme` binary into `build/`.
+- `make run`: Builds and runs the binary.
+- `make clean`: Removes the build binary and test databases.
+
+## Contributing
+
+Contributions are welcome! Please review the documentation in the `docs/` folder to understand the architecture and UI guidelines before submitting pull requests.
+
+## License
+
+MIT License.

cmd/hireme/main.go

@@ -6,12 +6,12 @@ import (
 	"os"
 	"path/filepath"
 
-	"github.com/hireme/hireme/internal/analyzer"
-	"github.com/hireme/hireme/internal/config"
-	"github.com/hireme/hireme/internal/db"
-	"github.com/hireme/hireme/internal/logger"
-	"github.com/hireme/hireme/internal/ui"
-	"github.com/hireme/hireme/internal/worker"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/analyzer"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/config"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/ui"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/worker"
 	"github.com/spf13/cobra"
 )
 
@@ -30,8 +30,7 @@ func newRootCmd() *cobra.Command {
 			logger.Log.Info("==========================================================")
 			logger.Log.Info("🚀 Starting Hireme (GJCC) Job Application Command Center")
 			logger.Log.Info("==========================================================")
-			logger.Log.Infof("[Config] Port: %s | DB Path: %s | Target Config: %s | Active LLM: %s | Log Level: %s",
-				cfg.Port, cfg.DBPath, cfg.TargetConfig, cfg.LLMProvider, cfg.LogLevel)
+			logger.Log.Infow("Starting Hireme", "component", "Config", "port", cfg.Port, "db_path", cfg.DBPath, "target_config", cfg.TargetConfig, "llm", cfg.LLMProvider, "log_level", cfg.LogLevel)
 
 			// Ensure DB directory exists
 			dbDir := filepath.Dir(cfg.DBPath)
@@ -48,7 +47,7 @@ func newRootCmd() *cobra.Command {
 
 			// Load targets from YAML config if database is empty
 			if err := database.LoadTargetsFromYAML(cfg.TargetConfig); err != nil {
-				logger.Log.Errorf("[Main] Target YAML loading status: %v", err)
+				logger.Log.Errorw("Target YAML loading failed", "component", "Main", "error", err)
 			}
 
 			// Initialize LLM Analyzer
@@ -60,11 +59,11 @@ func newRootCmd() *cobra.Command {
 			// Load filter configuration
 			filterCfg, err := config.LoadFilterConfig(cfg.FilterConfig)
 			if err != nil {
-				logger.Log.Fatalf("Failed to load filter config: %v", err)
+				logger.Log.Fatalw("Failed to load filter config", "component", "Main", "error", err)
 			}
 
 			// 4. Initialize Worker Pool
-			pool := worker.NewWorkerPool(database, llmAnalyzer, 3, filterCfg)
+			pool := worker.NewWorkerPool(database, llmAnalyzer, 3, cfg.AnalysisThreshold, filterCfg)
 			pool.Start() // Auto start the worker pool
 
 			// Setup UI Server & Routes
@@ -73,7 +72,7 @@ func newRootCmd() *cobra.Command {
 			server.RegisterRoutes(mux)
 
 			addr := fmt.Sprintf(":%s", cfg.Port)
-			logger.Log.Infof("✨ Hireme Command Center is live at http://localhost:%s", cfg.Port)
+			logger.Log.Infow("Hireme Command Center is live", "component", "Main", "url", fmt.Sprintf("http://localhost:%s", cfg.Port))
 			return http.ListenAndServe(addr, mux)
 		},
 	}

data/config/config.yaml

@@ -0,0 +1,13 @@
+# Hireme System Configuration
+
+server:
+  port: "8080"
+  log_level: "info"
+
+ai_trigger:
+  # Minimum keyword fit score from filters.yaml required to trigger deep LLM analysis (e.g. 25.0)
+  analysis_threshold: 25.0
+
+prompts:
+  system_prompt_path: "data/config/prompts/system_prompt.txt"
+  user_prompt_path: "data/config/prompts/user_prompt.txt"

data/config/filters.yaml

@@ -1,31 +1,103 @@
+# Hard exclusion keywords (immediate rejection)
 hard_keywords:
+  # Sponsorship / Clearance blockers
   - "no sponsorship"
+  - "without sponsorship"
   - "us citizen only"
+  - "u.s. citizen only"
+  - "active security clearance"
+  - "top secret/sci"
+  - "polygraph"
+  # Non-target employment types
   - "unpaid"
   - "volunteer"
+  - "internship"
+  - "co-op"
+  # Unrelated roles
+  - "sales representative"
+  - "account executive"
+  - "recruiter"
+  - "marketing manager"
 
+# Minimum score threshold to qualify for deep LLM analysis
+analysis_threshold: 25.0
+
+# Scored keywords with weights
 scoring_weights:
-  - keyword: golang
-    weight: 3.0
-  - keyword: go
-    weight: 2.0
-  - keyword: python
-    weight: 2.0
-  - keyword: kubernetes
-    weight: 3.0
-  - keyword: k8s
-    weight: 3.0
-  - keyword: docker
-    weight: 2.0
-  - keyword: aws
-    weight: 1.0
-  - keyword: gcp
-    weight: 1.0
-  - keyword: distributed systems
-    weight: 4.0
-  - keyword: backend
-    weight: 3.0
-  - keyword: ai
-    weight: -5.0
-  - keyword: machine learning
-    weight: -5.0
+  # Tier 1: Core Systems, Cloud & AI Infrastructure (High Priority)
+  - keyword: "kubernetes"
+    weight: 15.0
+  - keyword: "k8s"
+    weight: 15.0
+  - keyword: "golang"
+    weight: 15.0
+  - keyword: "go"
+    weight: 12.0
+  - keyword: "distributed systems"
+    weight: 15.0
+  - keyword: "ai runtime"
+    weight: 15.0
+  - keyword: "deep learning system"
+    weight: 15.0
+  - keyword: "machine learning infrastructure"
+    weight: 15.0
+  - keyword: "gke"
+    weight: 12.0
+  - keyword: "container runtime"
+    weight: 12.0
+  - keyword: "containerd"
+    weight: 10.0
+  - keyword: "c++"
+    weight: 10.0
+  - keyword: "linux kernel"
+    weight: 12.0
+  - keyword: "operating system"
+    weight: 10.0
+
+  # Tier 2: Cloud, Backend, Systems Programming (Medium Priority)
+  - keyword: "docker"
+    weight: 8.0
+  - keyword: "grpc"
+    weight: 8.0
+  - keyword: "protobuf"
+    weight: 6.0
+  - keyword: "microservices"
+    weight: 8.0
+  - keyword: "python"
+    weight: 6.0
+  - keyword: "cloud infrastructure"
+    weight: 10.0
+  - keyword: "virtualization"
+    weight: 8.0
+  - keyword: "hypervisor"
+    weight: 8.0
+  - keyword: "arm"
+    weight: 6.0
+  - keyword: "performance optimization"
+    weight: 8.0
+  - keyword: "low latency"
+    weight: 8.0
+
+  # Tier 3: Seniority & Qualifications
+  - keyword: "phd"
+    weight: 8.0
+  - keyword: "staff software engineer"
+    weight: 10.0
+  - keyword: "senior software engineer"
+    weight: 8.0
+  - keyword: "principal engineer"
+    weight: 10.0
+  - keyword: "tech lead"
+    weight: 6.0
+
+  # Tier 4: Negative/Irrelevant Focus Penalties
+  - keyword: "react native"
+    weight: -8.0
+  - keyword: "wordpress"
+    weight: -15.0
+  - keyword: "php"
+    weight: -10.0
+  - keyword: "salesforce"
+    weight: -10.0
+  - keyword: "frontend only"
+    weight: -10.0

data/config/prompts/system_prompt.txt

@@ -0,0 +1,35 @@
+You are an expert Silicon Valley technical recruiter and hiring manager evaluating job descriptions for a specific candidate.
+
+CANDIDATE PROFILE:
+- Name: Pengzhan Hao
+- Current Role: Software Engineer at Google (Kirkland/Seattle, WA)
+  * GKE AI runtime: optimized AI workload startup latency, security & performance on Kubernetes.
+  * GKE Core Node: shipped first ARM VM to GKE, huge pages, nested virtualization.
+- Education:
+  * Ph.D. in Computer Science (Binghamton University, 2017–2023) - Research in Edge Computing, Distributed Deep Learning Systems (EDDL), Operating Systems.
+  * M.S. in Computer Science (Binghamton University, GPA: 4.0).
+  * B.E. in Software Engineering (Beihang University).
+- Technical Stack: Golang, C/C++, Python, Kubernetes (K8s), GKE, Containerd, Docker, Linux Kernel/OS, Distributed Systems, gRPC, ProtoBuf, ARM, Xen Hypervisor, PyTorch/Dlib.
+- Target Roles & Seniority: Senior Software Engineer / Staff Software Engineer / AI Infrastructure / Cloud Systems / Low-level Platforms.
+
+EVALUATION CRITERIA:
+1. Score from 0.0 to 100.0 on overall match based on candidate's exact background.
+   - 80-100: Exceptional match (Core Go/C++/K8s/AI Infra/Distributed Systems/Google scale).
+   - 60-79: Solid match (Good general backend/cloud infrastructure fit).
+   - 40-59: Moderate match (Significant domain divergence or missing core skill).
+   - 0-39: Poor match (Non-tech, frontend-only, or strict clearance requirements).
+2. is_qualified: boolean (true if score >= 60.0 and no hard disqualifiers).
+3. summary: 1-2 sentence executive assessment of fit.
+4. matched_highlights: 3-5 concrete technical strengths/experiences matched.
+5. missing_highlights: 1-3 genuine gaps, missing requirements, or nice-to-haves.
+6. tailoring_suggestions: 2-3 specific action points for tailoring resume/talking points for this JD.
+
+You MUST respond strictly with a valid JSON object matching the requested schema:
+{
+  "ai_score": 85.0,
+  "is_qualified": true,
+  "summary": "...",
+  "matched_highlights": ["..."],
+  "missing_highlights": ["..."],
+  "tailoring_suggestions": ["..."]
+}

data/config/prompts/user_prompt.txt

@@ -0,0 +1,8 @@
+Analyze the following job description for candidate Pengzhan Hao.
+
+Job Title: {{.JobTitle}}
+
+Job Description:
+{{.JobContent}}
+
+Provide the structured evaluation adhering to the JSON schema.

data/config/targets.yaml

@@ -1,5 +1,65 @@
 targets:
-  - name: "Google Software Engineer (Seattle, MID Level)"
-    target_url: "https://www.google.com/about/careers/applications/jobs/results/?employment_type=FULL_TIME&q=%22Software%20Engineer%22&location=Seattle%2C%20WA%2C%20USA&target_level=MID"
-    platform: "google_jobs"
+  - name: Google Software Engineer (Seattle, MID Level)
+    target_url: https://www.google.com/about/careers/applications/jobs/results/?employment_type=FULL_TIME&q=%22Software%20Engineer%22&location=Seattle%2C%20WA%2C%20USA&target_level=MID
+    platform: google_jobs
+    is_active: true
+
+  - name: Apple Software Engineer (Seattle)
+    target_url: https://jobs.apple.com/en-us/search?location=seattle-SEA&key=Software%2520Engineer&team=apps-and-frameworks-SFTWR-AF+cloud-and-infrastructure-SFTWR-CLD+core-operating-systems-SFTWR-COS+devops-and-site-reliability-SFTWR-DSR+engineering-project-management-SFTWR-EPM+information-systems-and-technology-SFTWR-ISTECH+machine-learning-and-ai-SFTWR-MCHLN+security-and-privacy-SFTWR-SEC+software-quality-automation-and-tools-SFTWR-SQAT+wireless-software-SFTWR-WSFT+machine-learning-infrastructure-MLAI-MLI+deep-learning-and-reinforcement-learning-MLAI-DLRL+natural-language-processing-and-speech-technologies-MLAI-NLP+computer-vision-MLAI-CV+applied-research-MLAI-ARr
+    platform: apple
+    is_active: true
+
+  - name: Meta Software Engineer (Seattle)
+    target_url: https://www.metacareers.com/jobsearch/?offices[0]=Seattle%2C%20WA&roles[0]=Full%20time%20employment&teams[0]=Software%20Engineering
+    platform: meta
+    is_active: true
+
+  - name: Amazon / AWS Software Engineer (Seattle / Bellevue)
+    target_url: https://www.amazon.jobs/en/search.json?base_query=Software+Development+Engineer&loc_query=Seattle%2C+WA%2C+USA
+    platform: amazon
+    is_active: true
+
+  - name: Microsoft Software Engineer (Redmond / Bellevue / Seattle)
+    target_url: https://gcsservices.careers.microsoft.com/search/api/v1/search?q=Software+Engineer&lc=Redmond%2C+Washington%2C+United+States&l=en_us
+    platform: microsoft
+    is_active: true
+
+  - name: ByteDance / TikTok Software Engineer (Seattle / Bellevue)
+    target_url: https://jobs.bytedance.com/api/v1/search/job/posts?keyword=Software+Engineer
+    platform: bytedance
+    is_active: true
+
+  - name: NVIDIA Software Engineer (Redmond / Seattle)
+    target_url: https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite?q=Software+Engineer
+    platform: nvidia
+    is_active: true
+
+  - name: Netflix Software Engineer (Seattle / Remote)
+    target_url: https://jobs.netflix.com/api/search?q=Software+Engineer
+    platform: netflix
+    is_active: true
+
+  - name: Uber Software Engineer (Seattle)
+    target_url: https://www.uber.com/global/en/careers/search/?query=Software+Engineer
+    platform: uber
+    is_active: true
+
+  - name: LinkedIn Software Engineer (Seattle / Bellevue)
+    target_url: https://www.linkedin.com/jobs-guest/jobs/api/seeMoreJobPostings/search?keywords=Software+Engineer&location=Seattle%2C+Washington%2C+United+States
+    platform: linkedin
+    is_active: true
+
+  - name: Stripe Software Engineer (Seattle / Remote)
+    target_url: https://boards-api.greenhouse.io/v1/boards/stripe/jobs
+    platform: stripe
+    is_active: true
+
+  - name: Databricks Software Engineer (Bellevue / Seattle)
+    target_url: https://boards-api.greenhouse.io/v1/boards/databricks/jobs
+    platform: databricks
+    is_active: true
+
+  - name: Snowflake Software Engineer (Bellevue / Seattle)
+    target_url: https://boards-api.greenhouse.io/v1/boards/snowflake/jobs
+    platform: snowflake
     is_active: true

docs/architecture.md

@@ -0,0 +1,43 @@
+# System Architecture
+
+Hireme is designed as a single-binary, local-first web application. It combines a high-performance backend, an embedded database, background job processing, and a lightweight frontend into a cohesive, easy-to-deploy system.
+
+## Components
+
+### 1. Database Layer (`internal/db`)
+
+- **Technology:** SQLite3 via `github.com/mattn/go-sqlite3`.
+- **Mode:** WAL (Write-Ahead Logging) mode is enabled (`PRAGMA journal_mode=WAL;`) to support high concurrency between the HTTP server reads and background worker writes without "database is locked" errors.
+- **Pattern:** Centralized query methods attached to a `DB` struct (`queries.go`).
+
+### 2. Worker Pool & Task Queue (`internal/worker`, `internal/db/tasks.go`)
+
+- **Technology:** Custom goroutine worker pool pulling from a `tasks` SQLite table.
+- **Tasks:**
+  - `DISCOVER`: Initiates crawling on a target (e.g., Google Jobs search URL).
+  - `FETCH`: Fetches raw HTML for a specific job posting.
+  - `PROCESS`: Parses raw HTML, applies rule-based filtering (keywords, scoring), and enqueues for analysis if qualified.
+  - `ANALYZE`: Sends the parsed job to an LLM for final qualification scoring and tailoring tips.
+- **Concurrency:** Managed via context cancellation and `atomic.Bool` flags for safe pausing and resuming.
+
+### 3. Filters & Rules (`internal/filter`, `data/config/filters.yaml`)
+
+- A pipeline pattern is used to quickly reject jobs before doing expensive LLM API calls.
+- **Hard Keywords:** Immediate rejection if a keyword (e.g., "Clearance", "Staff") is found.
+- **Scoring Weights:** Basic substring counting for required technologies (e.g., "Go", "Python"). If the raw score falls below a threshold, the job is rejected.
+
+### 4. Crawlers (`pkg/crawler`)
+
+- Currently implements Google Jobs (`pkg/crawler/google`).
+- **Mechanism:** Parses embedded JSON (`AF_initDataCallback`) in the raw HTML response rather than fragile DOM traversal. This ensures stability against UI changes.
+
+### 5. UI Layer (`internal/ui`)
+
+- **Technology:** Standard Go HTTP server (`net/http`), Go Templates (`html/template`), HTMX, and Tailwind CSS.
+- **Philosophy:** Server-Side Rendered (SSR) HTML fragments returned directly to HTMX for seamless, single-page application (SPA) like updates without heavy JS payloads.
+- **Templates:** Statically compiled strings in `.go` files (`template_layout.go`, `template_jobs.go`) rather than external `.html` files to keep the binary fully self-contained.
+
+### 6. Logging (`internal/logger`)
+
+- **Technology:** `go.uber.org/zap` (Sugared Logger).
+- **Pattern:** Structured logging using `w` methods (`Infow`, `Errorw`, `Debugw`) to allow for strongly-typed key-value pairs (`"component", "API"`) instead of string formatting.

docs/ui-guidelines.md

@@ -0,0 +1,38 @@
+# UI Guidelines
+
+Hireme uses a modern, server-rendered UI stack that eliminates the need for heavy JavaScript frameworks (React, Vue, etc.) while maintaining a highly reactive, single-page application (SPA) feel.
+
+## Stack
+
+- **HTML / Go Templates:** Server-side rendering using `html/template`.
+- **HTMX (`hx-*`):** Used for all dynamic interactions, form submissions, and UI updates.
+- **Tailwind CSS:** Utility-first CSS framework for styling via CDN.
+
+## Architecture & Patterns
+
+### 1. Template Strings vs Files
+
+Templates are declared as `const` strings in `.go` files (e.g., `template_layout.go`, `template_jobs.go`) and collected in a `TemplateRegistry` in `templates.go`. This keeps the final build as a single executable without requiring external static asset folders.
+
+### 2. HTMX Updates
+
+Instead of returning JSON and re-rendering via JS, API endpoints return rendered HTML fragments.
+
+- **Example:** Clicking `+ Save` on a job POSTs to `/applications`. The handler updates the DB and returns the `template_jobs` HTML fragment, which HTMX swaps directly into the `#jobs-container` DOM element.
+
+### 3. Layout & Navigation (Tab Persistence)
+
+- The main UI uses a tabbed layout (Jobs, Apply, Targets).
+- Switching tabs is handled via lightweight Vanilla JS (`switchTab(id)`) which toggles `.hidden` classes.
+- Tab states are synced to the URL `#hash` so reloading the page preserves the active tab.
+
+### 4. Client-Side Filtering & Sorting
+
+- To prevent unnecessary server load for display formatting, table sorting and filtering (e.g., searching by keyword, or filtering by "Qualified") is done purely client-side using Vanilla JS.
+- Data attributes (`data-sort-val`, `data-qual`, `data-app`) are embedded in the `<tr>` elements during server-render to aid the JS filter logic.
+
+### 5. Design Aesthetics
+
+- **Dark Mode Default:** The UI uses Tailwind's `dark:` classes implicitly by having `class="dark"` on the root `<html>` tag.
+- **Micro-interactions:** Use hover states (`hover:bg-slate-800`), transitions (`transition-colors`), and subtle borders to create a premium feel.
+- **Indicators:** Use badges and color-coded text (Emerald for success/qualified, Rose for rejected/not qualified, Indigo for actions).

go.mod

@@ -1,9 +1,11 @@
-module github.com/hireme/hireme
+module git.pengzhan.dev/haopengzhan/hireme
 
 go 1.26
 
 require (
 	github.com/JohannesKaufmann/html-to-markdown/v2 v2.5.2
+	github.com/chromedp/cdproto v0.0.0-20260714215040-dc233986426f
+	github.com/chromedp/chromedp v0.16.0
 	github.com/mattn/go-sqlite3 v1.14.42
 	github.com/spf13/cobra v1.9.1
 	github.com/spf13/pflag v1.0.6
@@ -14,6 +16,12 @@ require (
 
 require (
 	github.com/JohannesKaufmann/dom v0.3.1 // indirect
+	github.com/chromedp/sysutil v1.1.0 // indirect
+	github.com/go-json-experiment/json v0.0.0-20260623181947-01eb4420fa68 // indirect
+	github.com/gobwas/httphead v0.1.0 // indirect
+	github.com/gobwas/pool v0.2.1 // indirect
+	github.com/gobwas/ws v1.4.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	go.uber.org/multierr v1.10.0 // indirect
+	golang.org/x/sys v0.47.0 // indirect
 )

go.sum

@@ -2,24 +2,59 @@ github.com/JohannesKaufmann/dom v0.3.1 h1:J16l9JAHWgkFPR3VIPbQ1gvS0cWab6laK1q7PF
 github.com/JohannesKaufmann/dom v0.3.1/go.mod h1:BZPkf8ZeYrBgABjwJn9iiKt8aiCtkxpHkevms+Yp2DE=
 github.com/JohannesKaufmann/html-to-markdown/v2 v2.5.2 h1:XFJZFWESIWlUEHHjzBuv8RvrtCWnSGlimEX17ysSDb8=
 github.com/JohannesKaufmann/html-to-markdown/v2 v2.5.2/go.mod h1:BHWO8lJzttJLqwuV8Rb1B3OG2OSzLbssZDI1FRg2eAA=
+github.com/chromedp/cdproto v0.0.0-20260714215040-dc233986426f h1:0Z1zcSLEmnj2c2CmJYBqewtS6pxhB39bNWUSEUAWjgk=
+github.com/chromedp/cdproto v0.0.0-20260714215040-dc233986426f/go.mod h1:RwFsSODCtFExll+GhHM6R92SARHR3Z3oipaxLHj46C0=
+github.com/chromedp/chromedp v0.16.0 h1:rOO4deOm4CbZgBCa8mD9g2rDyIoNs0BkgvNrlbp5ouk=
+github.com/chromedp/chromedp v0.16.0/go.mod h1:rbuGKFT1vMcFcFqKfPIO1GpX/N+2s8onm2qMxZLbU5U=
+github.com/chromedp/sysutil v1.1.0 h1:PUFNv5EcprjqXZD9nJb9b/c9ibAbxiYo4exNWZyipwM=
+github.com/chromedp/sysutil v1.1.0/go.mod h1:WiThHUdltqCNKGc4gaU50XgYjwjYIhKWoHGPTUfWTJ8=
 github.com/cpuguy83/go-md2man/v2 v2.0.6/go.mod h1:oOW0eioCTA6cOiMLiUPZOpcVxMig6NIQQ7OS05n1F4g=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/go-json-experiment/json v0.0.0-20260623181947-01eb4420fa68 h1:KZaTBSyshWX3MP5jukJcNSuXDQTO+rNpt0J564dX/eg=
+github.com/go-json-experiment/json v0.0.0-20260623181947-01eb4420fa68/go.mod h1:tphK2c80bpPhMOI4v6bIc2xWywPfbqi1Z06+RcrMkDg=
+github.com/gobwas/httphead v0.1.0 h1:exrUm0f4YX0L7EBwZHuCF4GDp8aJfVeBrlLQrs6NqWU=
+github.com/gobwas/httphead v0.1.0/go.mod h1:O/RXo79gxV8G+RqlR/otEwx4Q36zl9rqC5u12GKvMCM=
+github.com/gobwas/pool v0.2.1 h1:xfeeEhW7pwmX8nuLVlqbzVc7udMDrwetjEv+TZIz1og=
+github.com/gobwas/pool v0.2.1/go.mod h1:q8bcK0KcYlCgd9e7WYLm9LpyS+YeLd8JVDW6WezmKEw=
+github.com/gobwas/ws v1.4.0 h1:CTaoG1tojrh4ucGPcoJFiAQUAsEWekEWvLy7GsVNqGs=
+github.com/gobwas/ws v1.4.0/go.mod h1:G3gNqMNtPppf5XUz7O4shetPpcZ1VJ7zt18dlUeakrc=
 github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8=
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
+github.com/ledongthuc/pdf v0.0.0-20220302134840-0c2507a12d80 h1:6Yzfa6GP0rIo/kULo2bwGEkFvCePZ3qHDDTC3/J9Swo=
+github.com/ledongthuc/pdf v0.0.0-20220302134840-0c2507a12d80/go.mod h1:imJHygn/1yfhB7XSJJKlFZKl/J+dCPAknuiaGOshXAs=
 github.com/mattn/go-sqlite3 v1.14.42 h1:MigqEP4ZmHw3aIdIT7T+9TLa90Z6smwcthx+Azv4Cgo=
 github.com/mattn/go-sqlite3 v1.14.42/go.mod h1:pjEuOr8IwzLJP2MfGeTb0A35jauH+C2kbHKBr7yXKVQ=
+github.com/orisano/pixelmatch v0.0.0-20220722002657-fb0b55479cde h1:x0TT0RDC7UhAVbbWWBzr41ElhJx5tXPWkIHA2HWPRuw=
+github.com/orisano/pixelmatch v0.0.0-20220722002657-fb0b55479cde/go.mod h1:nZgzbfBr3hhjoZnS66nKrHmduYNpc34ny7RK4z5/HM0=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
+github.com/sebdah/goldie/v2 v2.8.0 h1:dZb9wR8q5++oplmEiJT+U/5KyotVD+HNGCAc5gNr8rc=
+github.com/sebdah/goldie/v2 v2.8.0/go.mod h1:oZ9fp0+se1eapSRjfYbsV/0Hqhbuu3bJVvKI/NNtssI=
+github.com/sergi/go-diff v1.4.0 h1:n/SP9D5ad1fORl+llWyN+D6qoUETXNZARKjyY2/KVCw=
+github.com/sergi/go-diff v1.4.0/go.mod h1:A0bzQcvG0E7Rwjx0REVgAGH58e96+X0MeOfepqsbeW4=
 github.com/spf13/cobra v1.9.1 h1:CXSaggrXdbHK9CF+8ywj8Amf7PBRmPCOJugH954Nnlo=
 github.com/spf13/cobra v1.9.1/go.mod h1:nDyEzZ8ogv936Cinf6g1RU9MRY64Ir93oCnqb9wxYW0=
 github.com/spf13/pflag v1.0.6 h1:jFzHGLGAlb3ruxLB8MhbI6A8+AQX/2eW4qeyNZXNp2o=
 github.com/spf13/pflag v1.0.6/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
+github.com/stretchr/testify v1.8.1 h1:w7B6lhMri9wdJUVmEZPGGhZzrYTPvgJArz7wNPgYKsk=
+github.com/stretchr/testify v1.8.1/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4=
+github.com/yuin/goldmark v1.8.2 h1:kEGpgqJXdgbkhcOgBxkC0X0PmoPG1ZyoZ117rDVp4zE=
+github.com/yuin/goldmark v1.8.2/go.mod h1:ip/1k0VRfGynBgxOz0yCqHrbZXhcjxyuS66Brc7iBKg=
+go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
+go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
 go.uber.org/multierr v1.10.0 h1:S0h4aNzvfcFsC3dRF1jLoaov7oRaKqRGC/pUEJ2yvPQ=
 go.uber.org/multierr v1.10.0/go.mod h1:20+QtiLqy0Nd6FdQB9TLXag12DsQkrbs3htMFfDN80Y=
 go.uber.org/zap v1.28.0 h1:IZzaP1Fv73/T/pBMLk4VutPl36uNC+OSUh3JLG3FIjo=
 go.uber.org/zap v1.28.0/go.mod h1:rDLpOi171uODNm/mxFcuYWxDsqWSAVkFdX4XojSKg/Q=
-golang.org/x/net v0.52.0 h1:He/TN1l0e4mmR3QqHMT2Xab3Aj3L9qjbhRm78/6jrW0=
-golang.org/x/net v0.52.0/go.mod h1:R1MAz7uMZxVMualyPXb+VaqGSa3LIaUqk0eEt3w36Sw=
+go.yaml.in/yaml/v3 v3.0.4 h1:tfq32ie2Jv2UxXFdLJdh3jXuOzWiL1fo0bu/FbuKpbc=
+go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
 golang.org/x/net v0.55.0 h1:bcvxaJn3e1U6InsFWt1JUq1aSjnRxLzT2rtD2KfkDF8=
 golang.org/x/net v0.55.0/go.mod h1:L5U2KuzuOe1lY7Z+aWVIKK6qEeJXnXV9yzGA+WCHJww=
+golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
+golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=

internal/analyzer/analyzer.go

@@ -5,10 +5,13 @@ import (
 	"fmt"
 	"strings"
 
-	"github.com/hireme/hireme/internal/config"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/config"
 )
 
 type AnalysisOutput struct {
+	AIScore              float64  `json:"ai_score"`
+	IsQualified          bool     `json:"is_qualified"`
+	Summary              string   `json:"summary"`
 	MatchedHighlights    []string `json:"matched_highlights"`
 	MissingHighlights    []string `json:"missing_highlights"`
 	TailoringSuggestions []string `json:"tailoring_suggestions"`
@@ -25,12 +28,12 @@ func NewAnalyzer(cfg *config.Config) (LLMAnalyzer, error) {
 		if cfg.GeminiAPIKey == "" {
 			return nil, fmt.Errorf("GEMINI_API_KEY is required for provider 'gemini'")
 		}
-		return NewGeminiAnalyzer(cfg.GeminiAPIKey), nil
+		return NewGeminiAnalyzer(cfg.GeminiAPIKey, cfg.SystemPrompt, cfg.UserPrompt), nil
 	case "deepseek":
 		if cfg.DeepSeekAPIKey == "" {
 			return nil, fmt.Errorf("DEEPSEEK_API_KEY is required for provider 'deepseek'")
 		}
-		return NewDeepSeekAnalyzer(cfg.DeepSeekAPIKey, cfg.DeepSeekBaseURL), nil
+		return NewDeepSeekAnalyzer(cfg.DeepSeekAPIKey, cfg.DeepSeekBaseURL, cfg.SystemPrompt, cfg.UserPrompt), nil
 	case "mock":
 		return NewMockAnalyzer(), nil
 	default:

internal/analyzer/analyzer_test.go

@@ -0,0 +1,50 @@
+package analyzer
+
+import (
+	"context"
+	"testing"
+)
+
+func TestCleanJSONResponse(t *testing.T) {
+	cases := []struct {
+		input    string
+		expected string
+	}{
+		{
+			input:    "```json\n{\"ai_score\": 90}\n```",
+			expected: "{\"ai_score\": 90}",
+		},
+		{
+			input:    "```\n{\"ai_score\": 85}\n```",
+			expected: "{\"ai_score\": 85}",
+		},
+		{
+			input:    "{\"ai_score\": 95}",
+			expected: "{\"ai_score\": 95}",
+		},
+	}
+
+	for _, c := range cases {
+		out := cleanJSONResponse(c.input)
+		if out != c.expected {
+			t.Errorf("cleanJSONResponse(%q) = %q, expected %q", c.input, out, c.expected)
+		}
+	}
+}
+
+func TestMockAnalyzer(t *testing.T) {
+	mock := NewMockAnalyzer()
+	res, err := mock.AnalyzeJD(context.Background(), "Senior AI Infrastructure Engineer", "Building GKE and K8s runtimes.")
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if res.AIScore <= 0 {
+		t.Errorf("expected positive AI score, got %f", res.AIScore)
+	}
+	if !res.IsQualified {
+		t.Errorf("expected IsQualified=true, got false")
+	}
+	if len(res.MatchedHighlights) == 0 {
+		t.Errorf("expected matched highlights to not be empty")
+	}
+}

internal/analyzer/deepseek.go

@@ -14,36 +14,46 @@ import (
 type DeepSeekAnalyzer struct {
 	apiKey             string
 	baseURL            string
+	systemPrompt       string
+	userPromptTemplate string
 	client             *http.Client
 }
 
-func NewDeepSeekAnalyzer(apiKey, baseURL string) *DeepSeekAnalyzer {
+func NewDeepSeekAnalyzer(apiKey, baseURL, systemPrompt, userPromptTemplate string) *DeepSeekAnalyzer {
 	if baseURL == "" {
 		baseURL = "https://api.deepseek.com"
 	}
 	return &DeepSeekAnalyzer{
 		apiKey:             apiKey,
 		baseURL:            strings.TrimRight(baseURL, "/"),
+		systemPrompt:       systemPrompt,
+		userPromptTemplate: userPromptTemplate,
 		client:             &http.Client{Timeout: 30 * time.Second},
 	}
 }
 
-func (d *DeepSeekAnalyzer) AnalyzeJD(ctx context.Context, jobTitle, rawContent string) (*AnalysisOutput, error) {
-	prompt := fmt.Sprintf(`You are an expert technical resume screener. Analyze the following job description for the title "%s".
-Provide a structured JSON output with 3 fields:
-- "matched_highlights": list of 3-5 core technical skills/requirements matched (e.g. ["Go (Golang)", "Microservices", "REST APIs"]).
-- "missing_highlights": list of 1-3 missing skills, experience gaps, or extra requirements (e.g. ["10+ yrs experience", "Kubernetes administration"]).
-- "tailoring_suggestions": list of 2-3 specific action points for tailoring a resume to this JD.
+func (d *DeepSeekAnalyzer) buildUserPrompt(jobTitle, rawContent string) string {
+	if d.userPromptTemplate == "" {
+		return fmt.Sprintf("Analyze the following job description for candidate Pengzhan Hao.\n\nJob Title: %s\n\nJob Description:\n%s\n\nProvide the structured evaluation adhering to the JSON schema.", jobTitle, rawContent)
+	}
+	res := strings.ReplaceAll(d.userPromptTemplate, "{{.JobTitle}}", jobTitle)
+	res = strings.ReplaceAll(res, "{{.JobContent}}", rawContent)
+	return res
+}
 
-Job Title: %s
-Job Content:
-%s`, jobTitle, jobTitle, rawContent)
+func (d *DeepSeekAnalyzer) AnalyzeJD(ctx context.Context, jobTitle, rawContent string) (*AnalysisOutput, error) {
+	sysPrompt := d.systemPrompt
+	if sysPrompt == "" {
+		sysPrompt = "You are an expert technical resume screener. You MUST respond with a valid JSON object matching {\"ai_score\": 80.0, \"is_qualified\": true, \"summary\": \"...\", \"matched_highlights\": [...], \"missing_highlights\": [...], \"tailoring_suggestions\": [...]}."
+	}
+
+	userPrompt := d.buildUserPrompt(jobTitle, rawContent)
 
 	reqBody := map[string]interface{}{
 		"model": "deepseek-chat",
 		"messages": []map[string]string{
-			{"role": "system", "content": "You are a professional hiring assistant. You MUST respond ONLY with valid JSON matching {\"matched_highlights\": [...], \"missing_highlights\": [...], \"tailoring_suggestions\": [...]}."},
-			{"role": "user", "content": prompt},
+			{"role": "system", "content": sysPrompt},
+			{"role": "user", "content": userPrompt},
 		},
 		"response_format": map[string]string{
 			"type": "json_object",
@@ -99,7 +109,28 @@ Job Content:
 
 	var output AnalysisOutput
 	if err := json.Unmarshal([]byte(rawJSON), &output); err != nil {
-		return nil, fmt.Errorf("failed to parse DeepSeek JSON output: %w", err)
+		return nil, fmt.Errorf("failed to parse DeepSeek JSON output: %w (content: %s)", err, rawJSON)
+	}
+
+	// Sanitize output bounds
+	if output.AIScore < 0 {
+		output.AIScore = 0
+	} else if output.AIScore > 100 {
+		output.AIScore = 100
+	}
+
+	if output.AIScore >= 60.0 {
+		output.IsQualified = true
+	}
+
+	if output.MatchedHighlights == nil {
+		output.MatchedHighlights = []string{}
+	}
+	if output.MissingHighlights == nil {
+		output.MissingHighlights = []string{}
+	}
+	if output.TailoringSuggestions == nil {
+		output.TailoringSuggestions = []string{}
 	}
 
 	return &output, nil

internal/analyzer/gemini.go

@@ -7,40 +7,49 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"strings"
 	"time"
 )
 
 type GeminiAnalyzer struct {
 	apiKey             string
+	systemPrompt       string
+	userPromptTemplate string
 	client             *http.Client
 }
 
-func NewGeminiAnalyzer(apiKey string) *GeminiAnalyzer {
+func NewGeminiAnalyzer(apiKey, systemPrompt, userPromptTemplate string) *GeminiAnalyzer {
 	return &GeminiAnalyzer{
 		apiKey:             apiKey,
+		systemPrompt:       systemPrompt,
+		userPromptTemplate: userPromptTemplate,
 		client:             &http.Client{Timeout: 30 * time.Second},
 	}
 }
 
+func (g *GeminiAnalyzer) buildUserPrompt(jobTitle, rawContent string) string {
+	if g.userPromptTemplate == "" {
+		return fmt.Sprintf("Analyze the following job description for candidate Pengzhan Hao.\n\nJob Title: %s\n\nJob Content:\n%s\n\nRespond with structured JSON.", jobTitle, rawContent)
+	}
+	res := strings.ReplaceAll(g.userPromptTemplate, "{{.JobTitle}}", jobTitle)
+	res = strings.ReplaceAll(res, "{{.JobContent}}", rawContent)
+	return res
+}
+
 func (g *GeminiAnalyzer) AnalyzeJD(ctx context.Context, jobTitle, rawContent string) (*AnalysisOutput, error) {
-	prompt := fmt.Sprintf(`You are an expert technical resume screener. Analyze the following job description for the title "%s".
-Provide a structured JSON output with 3 fields:
-- "matched_highlights": list of 3-5 core technical skills/requirements matched (e.g. ["Go (Golang)", "Microservices", "REST APIs"]).
-- "missing_highlights": list of 1-3 missing skills, experience gaps, or extra requirements (e.g. ["10+ yrs experience", "Kubernetes administration"]).
-- "tailoring_suggestions": list of 2-3 specific action points for tailoring a resume to this JD.
+	sysPrompt := g.systemPrompt
+	if sysPrompt == "" {
+		sysPrompt = "You are an expert technical resume screener. Respond STRICTLY with valid JSON matching {\"ai_score\": 85.0, \"is_qualified\": true, \"summary\": \"...\", \"matched_highlights\": [...], \"missing_highlights\": [...], \"tailoring_suggestions\": [...]}."
+	}
 
-Job Title: %s
-Job Content:
-%s
-
-Respond STRICTLY with valid JSON matching the format:
-{"matched_highlights": [...], "missing_highlights": [...], "tailoring_suggestions": [...]}`, jobTitle, jobTitle, rawContent)
+	userPrompt := g.buildUserPrompt(jobTitle, rawContent)
+	fullPrompt := fmt.Sprintf("%s\n\n%s", sysPrompt, userPrompt)
 
 	reqBody := map[string]interface{}{
 		"contents": []map[string]interface{}{
 			{
 				"parts": []map[string]string{
-					{"text": prompt},
+					{"text": fullPrompt},
 				},
 			},
 		},
@@ -102,5 +111,15 @@ Respond STRICTLY with valid JSON matching the format:
 		return nil, fmt.Errorf("failed to parse Gemini JSON output: %w", err)
 	}
 
+	if output.AIScore < 0 {
+		output.AIScore = 0
+	} else if output.AIScore > 100 {
+		output.AIScore = 100
+	}
+
+	if output.AIScore >= 60.0 {
+		output.IsQualified = true
+	}
+
 	return &output, nil
 }

internal/analyzer/mock.go

@@ -10,11 +10,13 @@ func NewMockAnalyzer() *MockAnalyzer {
 	return &MockAnalyzer{}
 }
 
-// AnalyzeJD returns empty slices for mock provider so no fake mock text is injected into side panel during debug
 func (m *MockAnalyzer) AnalyzeJD(ctx context.Context, jobTitle, rawContent string) (*AnalysisOutput, error) {
 	return &AnalysisOutput{
-		MatchedHighlights:    []string{},
+		AIScore:              75.0,
+		IsQualified:          true,
+		Summary:              "Mock analysis: relevant systems and distributed infrastructure role.",
+		MatchedHighlights:    []string{"Golang", "Distributed Systems", "Kubernetes"},
 		MissingHighlights:    []string{},
-		TailoringSuggestions: []string{},
+		TailoringSuggestions: []string{"Highlight Google GKE runtime experience."},
 	}, nil
 }

internal/config/config.go

@@ -5,18 +5,72 @@ import (
 	"strings"
 
 	"github.com/spf13/pflag"
+	"gopkg.in/yaml.v3"
 )
 
+type AppConfigFile struct {
+	Server struct {
+		Port     string `yaml:"port"`
+		LogLevel string `yaml:"log_level"`
+	} `yaml:"server"`
+	AITrigger struct {
+		AnalysisThreshold float64 `yaml:"analysis_threshold"`
+	} `yaml:"ai_trigger"`
+	Prompts struct {
+		SystemPromptPath string `yaml:"system_prompt_path"`
+		UserPromptPath   string `yaml:"user_prompt_path"`
+	} `yaml:"prompts"`
+}
+
 type Config struct {
 	Port              string
 	DBPath            string
 	TargetConfig      string
 	FilterConfig      string
+	AppConfig         string
 	LogLevel          string
 	LLMProvider       string // "gemini", "deepseek", "mock"
 	GeminiAPIKey      string
 	DeepSeekAPIKey    string
 	DeepSeekBaseURL   string
+	AnalysisThreshold float64
+	SystemPromptPath  string
+	UserPromptPath    string
+	SystemPrompt      string
+	UserPrompt        string
+}
+
+func init() {
+	LoadEnvFile(".env")
+}
+
+// LoadEnvFile parses a simple KEY=VALUE .env file if it exists without overriding existing env vars
+func LoadEnvFile(filenames ...string) {
+	if len(filenames) == 0 {
+		filenames = []string{".env"}
+	}
+	for _, filename := range filenames {
+		data, err := os.ReadFile(filename)
+		if err != nil {
+			continue
+		}
+		lines := strings.Split(string(data), "\n")
+		for _, line := range lines {
+			line = strings.TrimSpace(line)
+			if line == "" || strings.HasPrefix(line, "#") {
+				continue
+			}
+			parts := strings.SplitN(line, "=", 2)
+			if len(parts) == 2 {
+				key := strings.TrimSpace(parts[0])
+				val := strings.TrimSpace(parts[1])
+				val = strings.Trim(val, `"'`)
+				if os.Getenv(key) == "" {
+					os.Setenv(key, val)
+				}
+			}
+		}
+	}
 }
 
 func RegisterFlags(flags *pflag.FlagSet) *Config {
@@ -26,21 +80,71 @@ func RegisterFlags(flags *pflag.FlagSet) *Config {
 	flags.StringVarP(&cfg.DBPath, "db", "d", getEnvOrDefault("DB_PATH", "data/db/hireme.db"), "Path to database file")
 	flags.StringVarP(&cfg.TargetConfig, "target-config", "c", getEnvOrDefault("TARGET_CONFIG", "data/config/targets.yaml"), "Path to targets YAML file")
 	flags.StringVarP(&cfg.FilterConfig, "filter-config", "f", getEnvOrDefault("FILTER_CONFIG", "data/config/filters.yaml"), "Path to filters YAML file")
+	flags.StringVar(&cfg.AppConfig, "app-config", getEnvOrDefault("APP_CONFIG", "data/config/config.yaml"), "Path to app YAML config file")
 	flags.StringVar(&cfg.LogLevel, "log-level", getEnvOrDefault("LOG_LEVEL", "info"), "Logging level (trace, debug, info, error)")
 	flags.StringVarP(&cfg.LLMProvider, "llm-provider", "l", getEnvOrDefault("LLM_PROVIDER", ""), "Active LLM Provider (gemini, deepseek, mock)")
 	flags.StringVar(&cfg.GeminiAPIKey, "gemini-api-key", os.Getenv("GEMINI_API_KEY"), "Gemini API Key")
 	flags.StringVar(&cfg.DeepSeekAPIKey, "deepseek-api-key", os.Getenv("DEEPSEEK_API_KEY"), "DeepSeek API Key")
 	flags.StringVar(&cfg.DeepSeekBaseURL, "deepseek-base-url", getEnvOrDefault("DEEPSEEK_BASE_URL", "https://api.deepseek.com"), "DeepSeek Base URL")
+	flags.Float64Var(&cfg.AnalysisThreshold, "analysis-threshold", 25.0, "Threshold score from filter to trigger LLM analysis")
 
 	return cfg
 }
 
 func (c *Config) Normalize() {
+	// Re-check env vars in case .env was loaded after initial flag binding
+	if c.DeepSeekAPIKey == "" {
+		c.DeepSeekAPIKey = os.Getenv("DEEPSEEK_API_KEY")
+	}
+	if c.GeminiAPIKey == "" {
+		c.GeminiAPIKey = os.Getenv("GEMINI_API_KEY")
+	}
+
+	// Load data/config/config.yaml if present
+	if c.AppConfig != "" {
+		if data, err := os.ReadFile(c.AppConfig); err == nil {
+			var appCfg AppConfigFile
+			if err := yaml.Unmarshal(data, &appCfg); err == nil {
+				if appCfg.Server.Port != "" && c.Port == "8080" {
+					c.Port = appCfg.Server.Port
+				}
+				if appCfg.Server.LogLevel != "" && c.LogLevel == "info" {
+					c.LogLevel = appCfg.Server.LogLevel
+				}
+				if appCfg.AITrigger.AnalysisThreshold > 0 {
+					c.AnalysisThreshold = appCfg.AITrigger.AnalysisThreshold
+				}
+				if appCfg.Prompts.SystemPromptPath != "" {
+					c.SystemPromptPath = appCfg.Prompts.SystemPromptPath
+				}
+				if appCfg.Prompts.UserPromptPath != "" {
+					c.UserPromptPath = appCfg.Prompts.UserPromptPath
+				}
+			}
+		}
+	}
+
+	if c.SystemPromptPath == "" {
+		c.SystemPromptPath = "data/config/prompts/system_prompt.txt"
+	}
+	if c.UserPromptPath == "" {
+		c.UserPromptPath = "data/config/prompts/user_prompt.txt"
+	}
+
+	// Load prompt contents from files
+	if promptData, err := os.ReadFile(c.SystemPromptPath); err == nil {
+		c.SystemPrompt = strings.TrimSpace(string(promptData))
+	}
+	if userPromptData, err := os.ReadFile(c.UserPromptPath); err == nil {
+		c.UserPrompt = strings.TrimSpace(string(userPromptData))
+	}
+
+	// Determine LLM provider
 	if c.LLMProvider == "" {
-		if c.GeminiAPIKey != "" {
-			c.LLMProvider = "gemini"
-		} else if c.DeepSeekAPIKey != "" {
+		if c.DeepSeekAPIKey != "" {
 			c.LLMProvider = "deepseek"
+		} else if c.GeminiAPIKey != "" {
+			c.LLMProvider = "gemini"
 		} else {
 			c.LLMProvider = "mock"
 		}

internal/config/config_test.go

@@ -0,0 +1,46 @@
+package config
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestConfigLoadEnvAndAppConfig(t *testing.T) {
+	tempDir := t.TempDir()
+	envPath := filepath.Join(tempDir, ".env")
+	configYAMLPath := filepath.Join(tempDir, "config.yaml")
+
+	_ = os.WriteFile(envPath, []byte("TEST_SECRET_KEY=12345\nDEEPSEEK_API_KEY=test-sk-key\n"), 0644)
+	_ = os.WriteFile(configYAMLPath, []byte(`
+server:
+  port: "9090"
+  log_level: "debug"
+ai_trigger:
+  analysis_threshold: 35.0
+prompts:
+  system_prompt_path: "sys.txt"
+`), 0644)
+
+	LoadEnvFile(envPath)
+	if os.Getenv("TEST_SECRET_KEY") != "12345" {
+		t.Errorf("expected TEST_SECRET_KEY to be 12345, got %q", os.Getenv("TEST_SECRET_KEY"))
+	}
+
+	cfg := &Config{
+		Port:      "8080",
+		LogLevel:  "info",
+		AppConfig: configYAMLPath,
+	}
+	cfg.Normalize()
+
+	if cfg.Port != "9090" {
+		t.Errorf("expected port 9090 from config.yaml, got %s", cfg.Port)
+	}
+	if cfg.AnalysisThreshold != 35.0 {
+		t.Errorf("expected analysis threshold 35.0, got %f", cfg.AnalysisThreshold)
+	}
+	if cfg.LLMProvider != "deepseek" {
+		t.Errorf("expected LLMProvider to be deepseek, got %s", cfg.LLMProvider)
+	}
+}

internal/config/filters.go

@@ -4,12 +4,13 @@ import (
 	"fmt"
 	"os"
 
-	"github.com/hireme/hireme/internal/filter"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/filter"
 	"gopkg.in/yaml.v3"
 )
 
 type FilterConfig struct {
 	HardKeywords      []string               `yaml:"hard_keywords"`
+	AnalysisThreshold float64                `yaml:"analysis_threshold"`
 	ScoringWeights    []filter.KeywordWeight `yaml:"scoring_weights"`
 }
 
@@ -20,6 +21,8 @@ func LoadFilterConfig(path string) (*FilterConfig, error) {
 	}
 
 	var cfg FilterConfig
+	// Default threshold if not specified
+	cfg.AnalysisThreshold = 25.0
 	if err := yaml.Unmarshal(data, &cfg); err != nil {
 		return nil, fmt.Errorf("failed to unmarshal filter config: %w", err)
 	}

internal/crawler/crawlertest/crawlertest.go

@@ -0,0 +1,326 @@
+// Package crawlertest provides a standardized integration test harness for
+// validating that each SiteCrawler still functions against its real target URL.
+//
+// Each crawler package should have a TestXxx_LiveIntegration test that calls
+// crawlertest.RunLiveIntegration(). This tests three real-world failure modes:
+//
+//  1. DISCOVER STAGE - Can we still hit the search URL and get job listings?
+//     Detects: search API endpoint moved, query param format changed, auth required.
+//
+//  2. FETCH STAGE - Can we still fetch a single job detail page?
+//     Detects: detail URL pattern changed, page returns 403/404, rate limiting.
+//
+//  3. PARSE STAGE - Can we still extract structured data from the response?
+//     Detects: HTML/JSON schema changed, title/company/content extraction broken.
+package crawlertest
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"testing"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+)
+
+// LiveTestConfig defines what a live integration test expects from a crawler.
+type LiveTestConfig struct {
+	// PlatformName is the human-readable platform name (e.g., "Amazon", "Google").
+	PlatformName string
+
+	// Crawler is the SiteCrawler instance to test.
+	Crawler crawler.SiteCrawler
+
+	// TargetURL is the real search/listing URL to test discovery against.
+	TargetURL string
+
+	// ExpectedCanonicalPrefix is the prefix each CanonicalID should start with
+	// (e.g., "amazon:", "google:"). Empty string skips this check.
+	ExpectedCanonicalPrefix string
+
+	// ExpectedCompany is the expected Company field in FetchJobDetail results.
+	// Empty string skips this check.
+	ExpectedCompany string
+
+	// MinDiscoverCount is the minimum number of jobs DiscoverJobs should return.
+	// Defaults to 1 if not set.
+	MinDiscoverCount int
+
+	// MinContentLength is the minimum byte length of RawContent in FetchJobDetail.
+	// Defaults to 100 if not set.
+	MinContentLength int
+
+	// Timeout is the context timeout for the entire test. Defaults to 30s.
+	Timeout time.Duration
+}
+
+// RunLiveIntegration runs a standardized 3-stage integration test against real
+// targets. It gracefully skips on network errors (offline/sandbox) but FAILS
+// loudly when the crawler reaches the site but can't parse data correctly.
+//
+// This is the test that tells you "the site changed and our crawler is broken".
+//
+// Usage in each crawler package's _test.go:
+//
+//	func TestAmazonCrawler_LiveIntegration(t *testing.T) {
+//	    crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+//	        PlatformName:            "Amazon",
+//	        Crawler:                 NewAmazonCrawler(),
+//	        TargetURL:               "https://www.amazon.jobs/en/search.json?...",
+//	        ExpectedCanonicalPrefix: "amazon:",
+//	        ExpectedCompany:         "Amazon",
+//	    })
+//	}
+func RunLiveIntegration(t *testing.T, cfg LiveTestConfig) {
+	t.Helper()
+
+	// Apply defaults
+	if cfg.MinDiscoverCount <= 0 {
+		cfg.MinDiscoverCount = 1
+	}
+	if cfg.MinContentLength <= 0 {
+		cfg.MinContentLength = 100
+	}
+	if cfg.Timeout <= 0 {
+		cfg.Timeout = 30 * time.Second
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), cfg.Timeout)
+	defer cancel()
+
+	// =========================================================================
+	// STAGE 1: DISCOVER — Can we reach the search URL and get job listings?
+	// =========================================================================
+	t.Run("Stage1_Discover", func(t *testing.T) {
+		t.Logf("[%s] Discovering jobs from: %s", cfg.PlatformName, cfg.TargetURL)
+
+		discovered, err := cfg.Crawler.DiscoverJobs(ctx, cfg.TargetURL)
+		if err != nil {
+			if isNetworkError(err) {
+				t.Skipf("SKIP (offline): %v", err)
+			}
+			t.Fatalf("❌ DISCOVER FAILED: %v\n"+
+				"  → The search URL may have changed format or the API endpoint moved.\n"+
+				"  → Target URL: %s", err, cfg.TargetURL)
+		}
+
+		if len(discovered) == 0 {
+			t.Fatalf("❌ DISCOVER RETURNED 0 JOBS\n"+
+				"  → The search results page format may have changed.\n"+
+				"  → Check if the target URL still returns results in a browser.\n"+
+				"  → Target URL: %s", cfg.TargetURL)
+		}
+
+		if len(discovered) < cfg.MinDiscoverCount {
+			t.Errorf("⚠️  DISCOVER LOW COUNT: got %d jobs, expected at least %d\n"+
+				"  → The search API may be returning fewer results than expected.",
+				len(discovered), cfg.MinDiscoverCount)
+		}
+
+		t.Logf("✅ Discovered %d jobs", len(discovered))
+
+		// Validate first result structure
+		first := discovered[0]
+		if first.ApplyURL == "" {
+			t.Errorf("❌ First discovered job has empty ApplyURL")
+		}
+		if first.CanonicalID == "" {
+			t.Errorf("❌ First discovered job has empty CanonicalID")
+		}
+		if cfg.ExpectedCanonicalPrefix != "" && !strings.HasPrefix(first.CanonicalID, cfg.ExpectedCanonicalPrefix) {
+			t.Errorf("❌ CANONICAL ID FORMAT CHANGED: got %q, expected prefix %q\n"+
+				"  → The job ID extraction pattern may need updating.",
+				first.CanonicalID, cfg.ExpectedCanonicalPrefix)
+		}
+
+		t.Logf("   First job: ID=%s URL=%s", first.CanonicalID, first.ApplyURL)
+	})
+
+	// Re-discover for stage 2 (subtests get their own scope)
+	discovered, err := cfg.Crawler.DiscoverJobs(ctx, cfg.TargetURL)
+	if err != nil || len(discovered) == 0 {
+		// Stage 1 already reported the failure
+		return
+	}
+
+	// =========================================================================
+	// STAGE 2: FETCH — Can we fetch a job detail page?
+	// =========================================================================
+	firstJob := discovered[0]
+	var detail *crawler.ExtractedJob
+
+	t.Run("Stage2_Fetch", func(t *testing.T) {
+		t.Logf("[%s] Fetching job detail: %s", cfg.PlatformName, firstJob.ApplyURL)
+
+		var fetchErr error
+		detail, fetchErr = cfg.Crawler.FetchJobDetail(ctx, firstJob.ApplyURL)
+		if fetchErr != nil {
+			if isNetworkError(fetchErr) {
+				t.Skipf("SKIP (offline): %v", fetchErr)
+			}
+			t.Fatalf("❌ FETCH FAILED: %v\n"+
+				"  → The job detail URL pattern may have changed.\n"+
+				"  → The page may now require JavaScript rendering.\n"+
+				"  → Job URL: %s", fetchErr, firstJob.ApplyURL)
+		}
+
+		if detail == nil {
+			t.Fatalf("❌ FETCH RETURNED nil ExtractedJob")
+		}
+
+		t.Logf("✅ Fetched job detail successfully")
+	})
+
+	if detail == nil {
+		return
+	}
+
+	// =========================================================================
+	// STAGE 3: PARSE — Can we extract structured data from the response?
+	// =========================================================================
+	t.Run("Stage3_Parse", func(t *testing.T) {
+		t.Logf("[%s] Validating parsed fields", cfg.PlatformName)
+		failures := 0
+
+		// Title
+		if strings.TrimSpace(detail.Title) == "" {
+			t.Errorf("❌ PARSE: Title is empty\n"+
+				"  → The title extraction selector/pattern may have changed.\n"+
+				"  → Job URL: %s", firstJob.ApplyURL)
+			failures++
+		} else {
+			t.Logf("   Title:    %s", detail.Title)
+		}
+
+		// Company
+		if strings.TrimSpace(detail.Company) == "" {
+			t.Errorf("❌ PARSE: Company is empty\n"+
+				"  → The company extraction logic may have changed.")
+			failures++
+		} else {
+			t.Logf("   Company:  %s", detail.Company)
+		}
+
+		if cfg.ExpectedCompany != "" && !strings.Contains(detail.Company, cfg.ExpectedCompany) {
+			t.Errorf("⚠️  PARSE: Company mismatch: got %q, expected to contain %q\n"+
+				"  → The company name format may have changed.",
+				detail.Company, cfg.ExpectedCompany)
+		}
+
+		// Location
+		if strings.TrimSpace(detail.Location) == "" {
+			t.Errorf("⚠️  PARSE: Location is empty\n"+
+				"  → Location extraction may have changed.")
+		} else {
+			t.Logf("   Location: %s", detail.Location)
+		}
+
+		// CanonicalID
+		if detail.CanonicalID == "" {
+			t.Errorf("❌ PARSE: CanonicalID is empty")
+			failures++
+		}
+
+		// ApplyURL
+		if detail.ApplyURL == "" {
+			t.Errorf("❌ PARSE: ApplyURL is empty")
+			failures++
+		}
+
+		// RawContent (the critical one for downstream LLM analysis)
+		contentLen := len(detail.RawContent)
+		if contentLen == 0 {
+			t.Errorf("❌ PARSE: RawContent is EMPTY\n"+
+				"  → The content extraction is completely broken.\n"+
+				"  → The page structure or API response schema has changed.\n"+
+				"  → Job URL: %s", firstJob.ApplyURL)
+			failures++
+		} else if contentLen < cfg.MinContentLength {
+			t.Errorf("⚠️  PARSE: RawContent is too short (%d bytes, min %d)\n"+
+				"  → The content extraction may be partially broken.\n"+
+				"  → Job URL: %s", contentLen, cfg.MinContentLength, firstJob.ApplyURL)
+		} else {
+			t.Logf("   Content:  %d bytes", contentLen)
+		}
+
+		if failures == 0 {
+			t.Logf("✅ All parse checks passed")
+		} else {
+			t.Logf("❌ %d parse check(s) failed — crawler needs fixing", failures)
+		}
+	})
+}
+
+// isNetworkError checks if an error is a network/DNS/connectivity issue
+// (as opposed to a parsing or HTTP status error from a reachable server).
+func isNetworkError(err error) bool {
+	if err == nil {
+		return false
+	}
+	msg := err.Error()
+	networkPatterns := []string{
+		"no such host",
+		"dial tcp",
+		"connection refused",
+		"network is unreachable",
+		"i/o timeout",
+		"TLS handshake timeout",
+		"operation not permitted",
+		"connect: operation not permitted",
+	}
+	for _, pattern := range networkPatterns {
+		if strings.Contains(msg, pattern) {
+			return true
+		}
+	}
+	return false
+}
+
+// FormatHealthReport generates a summary table from multiple test results.
+// This can be called from a top-level aggregation test.
+func FormatHealthReport(results []HealthResult) string {
+	var sb strings.Builder
+	sb.WriteString(fmt.Sprintf("%-15s | %-10s | %-10s | %-10s | %s\n",
+		"Platform", "Discover", "Fetch", "Content", "Status"))
+	sb.WriteString(strings.Repeat("-", 70) + "\n")
+	for _, r := range results {
+		discoverStr := fmt.Sprintf("%d jobs", r.DiscoverCount)
+		fetchStr := "✓"
+		contentStr := fmt.Sprintf("%d B", r.ContentLength)
+		status := "✅ HEALTHY"
+
+		if r.Skipped {
+			status = "⏭ SKIPPED"
+			discoverStr = "-"
+			fetchStr = "-"
+			contentStr = "-"
+		} else if r.Error != nil {
+			status = "❌ BROKEN"
+			if r.DiscoverCount == 0 {
+				discoverStr = "FAIL"
+				fetchStr = "-"
+				contentStr = "-"
+			} else if r.ContentLength == 0 {
+				fetchStr = "FAIL"
+				contentStr = "-"
+			}
+		} else if r.DiscoverCount == 0 {
+			status = "⚠️  DEGRADED"
+			discoverStr = "0 ⚠️"
+		}
+
+		sb.WriteString(fmt.Sprintf("%-15s | %-10s | %-10s | %-10s | %s\n",
+			r.Platform, discoverStr, fetchStr, contentStr, status))
+	}
+	return sb.String()
+}
+
+// HealthResult is a structured result from a crawler health check.
+type HealthResult struct {
+	Platform      string
+	DiscoverCount int
+	ContentLength int
+	Error         error
+	Skipped       bool
+}

internal/db/db.go

@@ -3,8 +3,8 @@ package db
 import (
 	"database/sql"
 	"fmt"
-	"log"
 
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
 	_ "github.com/mattn/go-sqlite3"
 )
 
@@ -31,7 +31,7 @@ func InitDB(dbPath string) (*DB, error) {
 		return nil, fmt.Errorf("failed to run sqlite migrations: %w", err)
 	}
 
-	log.Printf("[DB] SQLite database initialized successfully at %s", dbPath)
+	logger.Log.Infow("SQLite database initialized successfully", "component", "DB", "dbPath", dbPath)
 	return database, nil
 }
 
@@ -55,14 +55,17 @@ func (db *DB) migrate() error {
 		location TEXT,
 		apply_url TEXT NOT NULL,
 		raw_content TEXT NOT NULL,
-		crawled_at DATETIME DEFAULT CURRENT_TIMESTAMP
+		crawled_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+		last_seen_at DATETIME DEFAULT CURRENT_TIMESTAMP
 	);
 
 	CREATE TABLE IF NOT EXISTS job_analyses (
 		id INTEGER PRIMARY KEY AUTOINCREMENT,
 		raw_job_id INTEGER UNIQUE NOT NULL REFERENCES raw_jobs(id) ON DELETE CASCADE,
+		filter_score REAL DEFAULT 0.0,
 		score REAL DEFAULT 0.0,
 		is_qualified BOOLEAN NOT NULL DEFAULT 0,
+		summary TEXT,
 		matched_highlights JSON,
 		missing_highlights JSON,
 		tailoring_suggestions JSON,
@@ -92,6 +95,72 @@ func (db *DB) migrate() error {
 	);
 	`
 
-	_, err := db.Exec(schema)
+	if _, err := db.Exec(schema); err != nil {
 		return err
 	}
+
+	// Dynamic migration for existing databases: check if last_seen_at exists in raw_jobs
+	var hasLastSeen bool
+	rows, err := db.Query(`PRAGMA table_info(raw_jobs)`)
+	if err == nil {
+		for rows.Next() {
+			var cid int
+			var name, ctype string
+			var notnull, pk int
+			var dfltValue sql.NullString
+			if err := rows.Scan(&cid, &name, &ctype, &notnull, &dfltValue, &pk); err == nil {
+				if name == "last_seen_at" {
+					hasLastSeen = true
+					break
+				}
+			}
+		}
+		_ = rows.Close()
+	}
+
+	if !hasLastSeen {
+		if _, err := db.Exec(`ALTER TABLE raw_jobs ADD COLUMN last_seen_at DATETIME`); err != nil {
+			logger.Log.Warnw("Failed to add last_seen_at column", "component", "DB", "error", err)
+		} else {
+			_, _ = db.Exec(`UPDATE raw_jobs SET last_seen_at = crawled_at WHERE last_seen_at IS NULL`)
+			_, _ = db.Exec(`UPDATE raw_jobs SET last_seen_at = CURRENT_TIMESTAMP WHERE last_seen_at IS NULL`)
+		}
+	}
+
+	// Dynamic migration for job_analyses: check if filter_score and summary exist
+	var hasFilterScore, hasSummary bool
+	rowsA, err := db.Query(`PRAGMA table_info(job_analyses)`)
+	if err == nil {
+		for rowsA.Next() {
+			var cid int
+			var name, ctype string
+			var notnull, pk int
+			var dfltValue sql.NullString
+			if err := rowsA.Scan(&cid, &name, &ctype, &notnull, &dfltValue, &pk); err == nil {
+				if name == "filter_score" {
+					hasFilterScore = true
+				}
+				if name == "summary" {
+					hasSummary = true
+				}
+			}
+		}
+		_ = rowsA.Close()
+	}
+
+	if !hasFilterScore {
+		if _, err := db.Exec(`ALTER TABLE job_analyses ADD COLUMN filter_score REAL DEFAULT 0.0`); err != nil {
+			logger.Log.Warnw("Failed to add filter_score column", "component", "DB", "error", err)
+		} else {
+			_, _ = db.Exec(`UPDATE job_analyses SET filter_score = score WHERE filter_score IS NULL OR filter_score = 0.0`)
+		}
+	}
+
+	if !hasSummary {
+		if _, err := db.Exec(`ALTER TABLE job_analyses ADD COLUMN summary TEXT`); err != nil {
+			logger.Log.Warnw("Failed to add summary column", "component", "DB", "error", err)
+		}
+	}
+
+	return nil
+}

internal/db/models.go

@@ -23,6 +23,7 @@ type RawJob struct {
 	ApplyURL    string    `json:"apply_url"`
 	RawContent  string    `json:"raw_content"`
 	CrawledAt   time.Time `json:"crawled_at"`
+	LastSeenAt  time.Time `json:"last_seen_at"`
 }
 
 type Task struct {
@@ -40,12 +41,14 @@ type Task struct {
 type JobAnalysis struct {
 	ID                   int64      `json:"id"`
 	RawJobID             int64      `json:"raw_job_id"`
+	FilterScore          float64    `json:"filter_score"`
 	Score                float64    `json:"score"`
 	IsQualified          bool       `json:"is_qualified"`
+	Summary              string     `json:"summary"`
 	MatchedHighlights    string     `json:"matched_highlights"`    // JSON array string
 	MissingHighlights    string     `json:"missing_highlights"`    // JSON array string
 	TailoringSuggestions string     `json:"tailoring_suggestions"` // JSON array string
-	Status               string     `json:"status"`               // 'PENDING', 'PARSED', 'FILTERED_OUT', 'FAILED'
+	Status               string     `json:"status"`               // 'PENDING', 'PARSED', 'FILTERED_OUT', 'REJECTED', 'FAILED'
 	AnalyzedAt           *time.Time `json:"analyzed_at"`
 }
 

internal/db/queries.go

@@ -2,6 +2,7 @@ package db
 
 import (
 	"database/sql"
+	"fmt"
 	"math"
 	"regexp"
 	"strings"
@@ -9,11 +10,11 @@ import (
 )
 
 // Target Queries
-func (db *DB) CreateTarget(name, targetURL, platform string) (*SearchTarget, error) {
-	query := `INSERT INTO search_targets (name, target_url, platform, is_active) VALUES (?, ?, ?, 1) RETURNING id, name, target_url, platform, is_active, last_crawled_at`
+func (db *DB) CreateTarget(name, targetURL, platform string, isActive bool) (*SearchTarget, error) {
+	query := `INSERT INTO search_targets (name, target_url, platform, is_active) VALUES (?, ?, ?, ?) RETURNING id, name, target_url, platform, is_active, last_crawled_at`
 	var target SearchTarget
 	var lastCrawled sql.NullTime
-	err := db.QueryRow(query, name, targetURL, platform).Scan(
+	err := db.QueryRow(query, name, targetURL, platform, isActive).Scan(
 		&target.ID, &target.Name, &target.TargetURL, &target.Platform, &target.IsActive, &lastCrawled,
 	)
 	if err != nil {
@@ -77,12 +78,98 @@ func (db *DB) ToggleTargetStatus(id int64) error {
 	return err
 }
 
+func (db *DB) SetTargetActive(id int64, isActive bool) error {
+	query := `UPDATE search_targets SET is_active = ? WHERE id = ?`
+	_, err := db.Exec(query, isActive, id)
+	return err
+}
+
 func (db *DB) UpdateTargetCrawledAt(id int64) error {
 	query := `UPDATE search_targets SET last_crawled_at = CURRENT_TIMESTAMP WHERE id = ?`
 	_, err := db.Exec(query, id)
 	return err
 }
 
+func (db *DB) DeleteTarget(id int64) (*SearchTarget, int64, error) {
+	tx, err := db.Begin()
+	if err != nil {
+		return nil, 0, err
+	}
+	defer func() { _ = tx.Rollback() }()
+
+	// 1. Get target info before deleting
+	var target SearchTarget
+	var lastCrawled sql.NullTime
+	err = tx.QueryRow(`SELECT id, name, target_url, platform, is_active, last_crawled_at FROM search_targets WHERE id = ?`, id).Scan(
+		&target.ID, &target.Name, &target.TargetURL, &target.Platform, &target.IsActive, &lastCrawled,
+	)
+	if err != nil {
+		return nil, 0, err
+	}
+	if lastCrawled.Valid {
+		target.LastCrawledAt = &lastCrawled.Time
+	}
+
+	// 2. Find all raw_jobs IDs associated with this target
+	rows, err := tx.Query(`SELECT id, canonical_id FROM raw_jobs WHERE target_id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+	var rawJobIDs []int64
+	var canonicalIDs []string
+	for rows.Next() {
+		var rID int64
+		var cID string
+		if err := rows.Scan(&rID, &cID); err != nil {
+			_ = rows.Close()
+			return nil, 0, err
+		}
+		rawJobIDs = append(rawJobIDs, rID)
+		canonicalIDs = append(canonicalIDs, cID)
+	}
+	_ = rows.Close()
+
+	// 3. Delete tasks referencing this target or its discovered jobs
+	targetIDPattern := fmt.Sprintf(`%%"target_id":%d%%`, id)
+	targetIDPatternStr := fmt.Sprintf(`%%"target_id": %d%%`, id)
+	_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ? OR payload LIKE ?`, targetIDPattern, targetIDPatternStr)
+
+	for _, cid := range canonicalIDs {
+		cPattern := fmt.Sprintf(`%%"canonical_id":"%s"%%`, cid)
+		_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ?`, cPattern)
+	}
+	for _, rID := range rawJobIDs {
+		rPattern := fmt.Sprintf(`%%"raw_job_id":%d%%`, rID)
+		rPatternStr := fmt.Sprintf(`%%"raw_job_id": %d%%`, rID)
+		_, _ = tx.Exec(`DELETE FROM tasks WHERE payload LIKE ? OR payload LIKE ?`, rPattern, rPatternStr)
+	}
+
+	// 4. Delete job_applications and job_analyses for these raw_jobs
+	for _, rID := range rawJobIDs {
+		_, _ = tx.Exec(`DELETE FROM job_applications WHERE raw_job_id = ?`, rID)
+		_, _ = tx.Exec(`DELETE FROM job_analyses WHERE raw_job_id = ?`, rID)
+	}
+
+	// 5. Delete raw_jobs for this target
+	res, err := tx.Exec(`DELETE FROM raw_jobs WHERE target_id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+	deletedJobsCount, _ := res.RowsAffected()
+
+	// 6. Delete search_target
+	_, err = tx.Exec(`DELETE FROM search_targets WHERE id = ?`, id)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	if err := tx.Commit(); err != nil {
+		return nil, 0, err
+	}
+
+	return &target, deletedJobsCount, nil
+}
+
 // Raw Jobs Queries & Vector Similarity Deduplication
 func (db *DB) InsertRawJob(canonicalID string, targetID *int64, title, company, location, applyURL, rawContent string) (*RawJob, bool, error) {
 	// Deduplication Check 1 & 2: Check canonical_id exist AND vector similarity >= 99%
@@ -104,15 +191,15 @@ func (db *DB) InsertRawJob(canonicalID string, targetID *int64, title, company,
 	}
 
 	query := `
-	INSERT INTO raw_jobs (canonical_id, target_id, title, company, location, apply_url, raw_content)
-	VALUES (?, ?, ?, ?, ?, ?, ?)
+	INSERT INTO raw_jobs (canonical_id, target_id, title, company, location, apply_url, raw_content, crawled_at, last_seen_at)
+	VALUES (?, ?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP, CURRENT_TIMESTAMP)
 	ON CONFLICT(canonical_id) DO NOTHING
-	RETURNING id, canonical_id, target_id, title, company, location, apply_url, raw_content, crawled_at
+	RETURNING id, canonical_id, target_id, title, company, location, apply_url, raw_content, crawled_at, last_seen_at
 	`
 	var job RawJob
 	var tID sql.NullInt64
 	err = db.QueryRow(query, canonicalID, targetID, title, company, location, applyURL, rawContent).Scan(
-		&job.ID, &job.CanonicalID, &tID, &job.Title, &job.Company, &job.Location, &job.ApplyURL, &job.RawContent, &job.CrawledAt,
+		&job.ID, &job.CanonicalID, &tID, &job.Title, &job.Company, &job.Location, &job.ApplyURL, &job.RawContent, &job.CrawledAt, &job.LastSeenAt,
 	)
 	if err == sql.ErrNoRows {
 		return nil, false, nil // Duplicate job
@@ -126,6 +213,125 @@ func (db *DB) InsertRawJob(canonicalID string, targetID *int64, title, company,
 	return &job, true, nil
 }
 
+func (db *DB) TouchJobLastSeen(canonicalID string) error {
+	query := `UPDATE raw_jobs SET last_seen_at = CURRENT_TIMESTAMP WHERE canonical_id = ?`
+	_, err := db.Exec(query, canonicalID)
+	return err
+}
+
+func (db *DB) TouchJobLastSeenByID(rawJobID int64) error {
+	query := `UPDATE raw_jobs SET last_seen_at = CURRENT_TIMESTAMP WHERE id = ?`
+	_, err := db.Exec(query, rawJobID)
+	return err
+}
+
+func (db *DB) ExpireJob(rawJobID int64, reason string) error {
+	if reason == "" {
+		reason = "Marked expired"
+	}
+	query := `
+	INSERT INTO job_applications (raw_job_id, status, notes, updated_at)
+	VALUES (?, 'EXPIRED', ?, CURRENT_TIMESTAMP)
+	ON CONFLICT(raw_job_id) DO UPDATE SET 
+		status='EXPIRED',
+		notes=COALESCE(job_applications.notes || ' | ', '') || excluded.notes,
+		updated_at=CURRENT_TIMESTAMP
+	`
+	_, err := db.Exec(query, rawJobID, reason)
+	return err
+}
+
+func (db *DB) EnqueueExpireTasks(days int, preserveActivePipeline bool) (int, error) {
+	tx, err := db.Begin()
+	if err != nil {
+		return 0, err
+	}
+	defer func() { _ = tx.Rollback() }()
+
+	var query string
+	if preserveActivePipeline {
+		query = `
+		SELECT r.id FROM raw_jobs r
+		LEFT JOIN job_applications app ON r.id = app.raw_job_id
+		WHERE r.last_seen_at < datetime('now', '-' || ? || ' days')
+		  AND (app.status IS NULL OR app.status NOT IN ('APPLIED', 'INTERVIEW', 'OFFER', 'EXPIRED', 'OPTED_OUT'))
+		`
+	} else {
+		query = `
+		SELECT r.id FROM raw_jobs r
+		LEFT JOIN job_applications app ON r.id = app.raw_job_id
+		WHERE r.last_seen_at < datetime('now', '-' || ? || ' days')
+		  AND (app.status IS NULL OR app.status NOT IN ('EXPIRED', 'OPTED_OUT'))
+		`
+	}
+
+	rows, err := tx.Query(query, days)
+	if err != nil {
+		return 0, err
+	}
+
+	var rawJobIDs []int64
+	for rows.Next() {
+		var id int64
+		if err := rows.Scan(&id); err != nil {
+			_ = rows.Close()
+			return 0, err
+		}
+		rawJobIDs = append(rawJobIDs, id)
+	}
+	_ = rows.Close()
+
+	for _, id := range rawJobIDs {
+		payloadJSON := fmt.Sprintf(`{"raw_job_id":%d,"reason":"Direct expire: not seen on target for %d+ days"}`, id, days)
+		if _, err := tx.Exec(`INSERT INTO tasks (task_type, payload, status) VALUES ('EXPIRE_JOB', ?, 'PENDING')`, payloadJSON); err != nil {
+			return 0, err
+		}
+	}
+
+	return len(rawJobIDs), tx.Commit()
+}
+
+func (db *DB) GetJobMaintenanceStats() (map[string]int, error) {
+	stats := make(map[string]int)
+
+	// Total raw jobs
+	var totalJobs int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM raw_jobs`).Scan(&totalJobs)
+	stats["total_jobs"] = totalJobs
+
+	// Active <= 7 days
+	var active7d int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM raw_jobs WHERE last_seen_at >= datetime('now', '-7 days')`).Scan(&active7d)
+	stats["active_7d"] = active7d
+
+	// Stale > 7 days
+	var stale7d int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM raw_jobs WHERE last_seen_at < datetime('now', '-7 days')`).Scan(&stale7d)
+	stats["stale_7d"] = stale7d
+
+	// Stale > 30 days
+	var stale30d int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM raw_jobs WHERE last_seen_at < datetime('now', '-30 days')`).Scan(&stale30d)
+	stats["stale_30d"] = stale30d
+
+	// Expired count
+	var expiredCount int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM job_applications WHERE status = 'EXPIRED'`).Scan(&expiredCount)
+	stats["expired_count"] = expiredCount
+
+	// Opted out count
+	var optedOutCount int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM job_applications WHERE status = 'OPTED_OUT'`).Scan(&optedOutCount)
+	stats["opted_out_count"] = optedOutCount
+
+	// In pipeline count
+	var inPipeline int
+	_ = db.QueryRow(`SELECT COUNT(*) FROM job_applications WHERE status IN ('SAVED', 'APPLIED', 'INTERVIEW', 'OFFER')`).Scan(&inPipeline)
+	stats["in_pipeline"] = inPipeline
+
+	return stats, nil
+}
+
 func (db *DB) CanonicalIDExists(canonicalID string) (bool, error) {
 	query := `SELECT EXISTS(SELECT 1 FROM raw_jobs WHERE canonical_id = ?)`
 	var exists bool
@@ -207,52 +413,60 @@ func tokenizeToVector(text string) map[string]int {
 	return freqMap
 }
 
-
-
 // Job Analyses Queries
-func (db *DB) CreateJobAnalysis(rawJobID int64, score float64, isQualified bool, status string) error {
+func (db *DB) CreateJobAnalysis(rawJobID int64, filterScore, score float64, isQualified bool, status string) error {
 	query := `
-	INSERT INTO job_analyses (raw_job_id, score, is_qualified, status)
-	VALUES (?, ?, ?, ?)
-	ON CONFLICT(raw_job_id) DO UPDATE SET score=excluded.score, is_qualified=excluded.is_qualified, status=excluded.status
+	INSERT INTO job_analyses (raw_job_id, filter_score, score, is_qualified, status)
+	VALUES (?, ?, ?, ?, ?)
+	ON CONFLICT(raw_job_id) DO UPDATE SET filter_score=excluded.filter_score, score=excluded.score, is_qualified=excluded.is_qualified, status=excluded.status
 	`
-	_, err := db.Exec(query, rawJobID, score, isQualified, status)
+	_, err := db.Exec(query, rawJobID, filterScore, score, isQualified, status)
 	return err
 }
 
-func (db *DB) CreateJobAnalysisFull(rawJobID int64, score float64, isQualified bool, matched, missing, tailoring, status string) error {
+func (db *DB) CreateJobAnalysisFull(rawJobID int64, filterScore, score float64, isQualified bool, summary, matched, missing, tailoring, status string) error {
 	query := `
-		INSERT INTO job_analyses (raw_job_id, score, is_qualified, matched_highlights, missing_highlights, tailoring_suggestions, status, analyzed_at) 
-		VALUES (?, ?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP)
+		INSERT INTO job_analyses (raw_job_id, filter_score, score, is_qualified, summary, matched_highlights, missing_highlights, tailoring_suggestions, status, analyzed_at) 
+		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, CURRENT_TIMESTAMP)
+		ON CONFLICT(raw_job_id) DO UPDATE SET 
+			filter_score=excluded.filter_score,
+			score=excluded.score,
+			is_qualified=excluded.is_qualified,
+			summary=excluded.summary,
+			matched_highlights=excluded.matched_highlights,
+			missing_highlights=excluded.missing_highlights,
+			tailoring_suggestions=excluded.tailoring_suggestions,
+			status=excluded.status,
+			analyzed_at=CURRENT_TIMESTAMP
 	`
-	_, err := db.Exec(query, rawJobID, score, isQualified, matched, missing, tailoring, status)
+	_, err := db.Exec(query, rawJobID, filterScore, score, isQualified, summary, matched, missing, tailoring, status)
 	return err
 }
 
-func (db *DB) UpdateJobAnalysisFull(rawJobID int64, score float64, isQualified bool, matchedJSON, missingJSON, suggestionsJSON, status string) error {
+func (db *DB) UpdateJobAnalysisFull(rawJobID int64, filterScore, score float64, isQualified bool, summary, matchedJSON, missingJSON, suggestionsJSON, status string) error {
 	query := `
 	UPDATE job_analyses 
-	SET score = ?, is_qualified = ?, matched_highlights = ?, missing_highlights = ?, tailoring_suggestions = ?, status = ?, analyzed_at = CURRENT_TIMESTAMP
+	SET filter_score = ?, score = ?, is_qualified = ?, summary = ?, matched_highlights = ?, missing_highlights = ?, tailoring_suggestions = ?, status = ?, analyzed_at = CURRENT_TIMESTAMP
 	WHERE raw_job_id = ?
 	`
-	_, err := db.Exec(query, score, isQualified, matchedJSON, missingJSON, suggestionsJSON, status, rawJobID)
+	_, err := db.Exec(query, filterScore, score, isQualified, summary, matchedJSON, missingJSON, suggestionsJSON, status, rawJobID)
 	return err
 }
 
-func (db *DB) UpdateJobAnalysisLLM(rawJobID int64, matchedJSON, missingJSON, suggestionsJSON string) error {
+func (db *DB) UpdateJobAnalysisLLM(rawJobID int64, aiScore float64, isQualified bool, summary, matchedJSON, missingJSON, suggestionsJSON string) error {
 	query := `
 	UPDATE job_analyses 
-	SET matched_highlights = ?, missing_highlights = ?, tailoring_suggestions = ?, status = 'PARSED', analyzed_at = CURRENT_TIMESTAMP
+	SET score = ?, is_qualified = ?, summary = ?, matched_highlights = ?, missing_highlights = ?, tailoring_suggestions = ?, status = 'PARSED', analyzed_at = CURRENT_TIMESTAMP
 	WHERE raw_job_id = ?
 	`
-	_, err := db.Exec(query, matchedJSON, missingJSON, suggestionsJSON, rawJobID)
+	_, err := db.Exec(query, aiScore, isQualified, summary, matchedJSON, missingJSON, suggestionsJSON, rawJobID)
 	return err
 }
 
 func (db *DB) GetPendingQualifiedAnalyses() ([]JobWithAnalysis, error) {
 	query := `
-	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at,
-	       a.id, a.score, a.is_qualified, COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), a.status, a.analyzed_at
+	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at, r.last_seen_at,
+	       a.id, COALESCE(a.filter_score, 0), COALESCE(a.score, 0), a.is_qualified, COALESCE(a.summary, ''), COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), a.status, a.analyzed_at
 	FROM raw_jobs r
 	JOIN job_analyses a ON r.id = a.raw_job_id
 	WHERE a.is_qualified = 1 AND a.status = 'PENDING'
@@ -268,14 +482,14 @@ func (db *DB) GetPendingQualifiedAnalyses() ([]JobWithAnalysis, error) {
 		var item JobWithAnalysis
 		var tID sql.NullInt64
 		var aID int64
-		var score float64
+		var filterScore, score float64
 		var isQualified bool
-		var matched, missing, suggestions, aStatus string
+		var summary, matched, missing, suggestions, aStatus string
 		var analyzedAt sql.NullTime
 
 		err := rows.Scan(
-			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt,
-			&aID, &score, &isQualified, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
+			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt, &item.LastSeenAt,
+			&aID, &filterScore, &score, &isQualified, &summary, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
 		)
 		if err != nil {
 			return nil, err
@@ -286,8 +500,10 @@ func (db *DB) GetPendingQualifiedAnalyses() ([]JobWithAnalysis, error) {
 		analysis := &JobAnalysis{
 			ID:                   aID,
 			RawJobID:             item.ID,
+			FilterScore:          filterScore,
 			Score:                score,
 			IsQualified:          isQualified,
+			Summary:              summary,
 			MatchedHighlights:    matched,
 			MissingHighlights:    missing,
 			TailoringSuggestions: suggestions,
@@ -304,13 +520,13 @@ func (db *DB) GetPendingQualifiedAnalyses() ([]JobWithAnalysis, error) {
 
 func (db *DB) GetJobsWithAnalysis() ([]JobWithAnalysis, error) {
 	query := `
-	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at,
-	       a.id, a.score, a.is_qualified, COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
+	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at, r.last_seen_at,
+	       a.id, COALESCE(a.filter_score, 0), COALESCE(a.score, 0), a.is_qualified, COALESCE(a.summary, ''), COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
 	       app.id, app.status, COALESCE(app.notes, ''), app.applied_at, app.updated_at
 	FROM raw_jobs r
 	LEFT JOIN job_analyses a ON r.id = a.raw_job_id
 	LEFT JOIN job_applications app ON r.id = app.raw_job_id
-	ORDER BY COALESCE(a.score, 0) DESC, r.crawled_at DESC
+	ORDER BY COALESCE(a.score, a.filter_score, 0) DESC, r.crawled_at DESC
 	`
 	rows, err := db.Query(query)
 	if err != nil {
@@ -323,17 +539,17 @@ func (db *DB) GetJobsWithAnalysis() ([]JobWithAnalysis, error) {
 		var item JobWithAnalysis
 		var tID sql.NullInt64
 		var aID, appID sql.NullInt64
-		var score sql.NullFloat64
+		var filterScore, score sql.NullFloat64
 		var isQualified sql.NullBool
-		var matched, missing, suggestions, aStatus sql.NullString
+		var summary, matched, missing, suggestions, aStatus sql.NullString
 		var analyzedAt sql.NullTime
 		var appStatus, notes sql.NullString
 		var appliedAt sql.NullTime
 		var updatedAt sql.NullTime
 
 		err := rows.Scan(
-			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt,
-			&aID, &score, &isQualified, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
+			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt, &item.LastSeenAt,
+			&aID, &filterScore, &score, &isQualified, &summary, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
 			&appID, &appStatus, &notes, &appliedAt, &updatedAt,
 		)
 		if err != nil {
@@ -346,8 +562,10 @@ func (db *DB) GetJobsWithAnalysis() ([]JobWithAnalysis, error) {
 			analysis := &JobAnalysis{
 				ID:                   aID.Int64,
 				RawJobID:             item.ID,
+				FilterScore:          filterScore.Float64,
 				Score:                score.Float64,
 				IsQualified:          isQualified.Bool,
+				Summary:              summary.String,
 				MatchedHighlights:    matched.String,
 				MissingHighlights:    missing.String,
 				TailoringSuggestions: suggestions.String,
@@ -378,8 +596,8 @@ func (db *DB) GetJobsWithAnalysis() ([]JobWithAnalysis, error) {
 
 func (db *DB) GetJobByID(rawJobID int64) (*JobWithAnalysis, error) {
 	query := `
-	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at,
-	       a.id, a.score, a.is_qualified, COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
+	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at, r.last_seen_at,
+	       a.id, COALESCE(a.filter_score, 0), COALESCE(a.score, 0), a.is_qualified, COALESCE(a.summary, ''), COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
 	       app.id, app.status, COALESCE(app.notes, ''), app.applied_at, app.updated_at
 	FROM raw_jobs r
 	LEFT JOIN job_analyses a ON r.id = a.raw_job_id
@@ -390,17 +608,17 @@ func (db *DB) GetJobByID(rawJobID int64) (*JobWithAnalysis, error) {
 	var item JobWithAnalysis
 	var tID sql.NullInt64
 	var aID, appID sql.NullInt64
-	var score sql.NullFloat64
+	var filterScore, score sql.NullFloat64
 	var isQualified sql.NullBool
-	var matched, missing, suggestions, aStatus sql.NullString
+	var summary, matched, missing, suggestions, aStatus sql.NullString
 	var analyzedAt sql.NullTime
 	var appStatus, notes sql.NullString
 	var appliedAt sql.NullTime
 	var updatedAt sql.NullTime
 
 	err := row.Scan(
-		&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt,
-		&aID, &score, &isQualified, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
+		&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt, &item.LastSeenAt,
+		&aID, &filterScore, &score, &isQualified, &summary, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
 		&appID, &appStatus, &notes, &appliedAt, &updatedAt,
 	)
 	if err != nil {
@@ -413,8 +631,10 @@ func (db *DB) GetJobByID(rawJobID int64) (*JobWithAnalysis, error) {
 		analysis := &JobAnalysis{
 			ID:                   aID.Int64,
 			RawJobID:             item.ID,
+			FilterScore:          filterScore.Float64,
 			Score:                score.Float64,
 			IsQualified:          isQualified.Bool,
+			Summary:              summary.String,
 			MatchedHighlights:    matched.String,
 			MissingHighlights:    missing.String,
 			TailoringSuggestions: suggestions.String,
@@ -475,8 +695,8 @@ func (db *DB) CreateOrUpdateApplication(rawJobID int64, status, notes string) (*
 
 func (db *DB) GetApplications() ([]JobWithAnalysis, error) {
 	query := `
-	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at,
-	       a.id, a.score, a.is_qualified, COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
+	SELECT r.id, r.canonical_id, r.target_id, r.title, r.company, r.location, r.apply_url, r.raw_content, r.crawled_at, r.last_seen_at,
+	       a.id, COALESCE(a.filter_score, 0), COALESCE(a.score, 0), a.is_qualified, COALESCE(a.summary, ''), COALESCE(a.matched_highlights, '[]'), COALESCE(a.missing_highlights, '[]'), COALESCE(a.tailoring_suggestions, '[]'), COALESCE(a.status, 'PENDING'), a.analyzed_at,
 	       app.id, app.status, COALESCE(app.notes, ''), app.applied_at, app.updated_at
 	FROM job_applications app
 	JOIN raw_jobs r ON app.raw_job_id = r.id
@@ -494,17 +714,17 @@ func (db *DB) GetApplications() ([]JobWithAnalysis, error) {
 		var item JobWithAnalysis
 		var tID sql.NullInt64
 		var aID, appID sql.NullInt64
-		var score sql.NullFloat64
+		var filterScore, score sql.NullFloat64
 		var isQualified sql.NullBool
-		var matched, missing, suggestions, aStatus sql.NullString
+		var summary, matched, missing, suggestions, aStatus sql.NullString
 		var analyzedAt sql.NullTime
 		var appStatus, notes sql.NullString
 		var appliedAt sql.NullTime
 		var updatedAt time.Time
 
 		err := rows.Scan(
-			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt,
-			&aID, &score, &isQualified, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
+			&item.ID, &item.CanonicalID, &tID, &item.Title, &item.Company, &item.Location, &item.ApplyURL, &item.RawContent, &item.CrawledAt, &item.LastSeenAt,
+			&aID, &filterScore, &score, &isQualified, &summary, &matched, &missing, &suggestions, &aStatus, &analyzedAt,
 			&appID, &appStatus, &notes, &appliedAt, &updatedAt,
 		)
 		if err != nil {
@@ -517,8 +737,10 @@ func (db *DB) GetApplications() ([]JobWithAnalysis, error) {
 			analysis := &JobAnalysis{
 				ID:                   aID.Int64,
 				RawJobID:             item.ID,
+				FilterScore:          filterScore.Float64,
 				Score:                score.Float64,
 				IsQualified:          isQualified.Bool,
+				Summary:              summary.String,
 				MatchedHighlights:    matched.String,
 				MissingHighlights:    missing.String,
 				TailoringSuggestions: suggestions.String,

internal/db/queries_test.go

@@ -1,7 +1,9 @@
 package db
 
 import (
+	"os"
 	"path/filepath"
+	"strings"
 	"testing"
 )
 
@@ -53,3 +55,245 @@ func TestVectorSimilarityDeduplication(t *testing.T) {
 		t.Errorf("first job mismatch")
 	}
 }
+
+func TestDeleteTargetCascade(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_delete_target.db")
+
+	database, err := InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	// 1. Create two search targets
+	target1, err := database.CreateTarget("Target 1", "https://example.com/target1", "google_jobs", true)
+	if err != nil {
+		t.Fatalf("failed to create target 1: %v", err)
+	}
+	target2, err := database.CreateTarget("Target 2", "https://example.com/target2", "apple", true)
+	if err != nil {
+		t.Fatalf("failed to create target 2: %v", err)
+	}
+
+	// 2. Insert raw jobs for target 1 and target 2
+	job1, _, err := database.InsertRawJob("canon-t1-j1", &target1.ID, "Job 1", "Google", "Seattle", "https://google.com/1", "Software Engineer Go")
+	if err != nil {
+		t.Fatalf("failed to insert job 1: %v", err)
+	}
+	job2, _, err := database.InsertRawJob("canon-t1-j2", &target1.ID, "Job 2", "Google", "Seattle", "https://google.com/2", "Software Engineer Python")
+	if err != nil {
+		t.Fatalf("failed to insert job 2: %v", err)
+	}
+	job3, _, err := database.InsertRawJob("canon-t2-j1", &target2.ID, "Job 3", "Apple", "Seattle", "https://apple.com/1", "Software Engineer iOS")
+	if err != nil {
+		t.Fatalf("failed to insert job 3: %v", err)
+	}
+
+	// 3. Create analyses and applications for target 1 jobs and target 2 jobs
+	_ = database.CreateJobAnalysis(job1.ID, 85.0, 85.0, true, "DONE")
+	_ = database.CreateJobAnalysis(job2.ID, 40.0, 40.0, false, "DONE")
+	_ = database.CreateJobAnalysis(job3.ID, 90.0, 90.0, true, "DONE")
+
+	_, _ = database.CreateOrUpdateApplication(job1.ID, "APPLIED", "Sent resume")
+	_, _ = database.CreateOrUpdateApplication(job3.ID, "SAVED", "Review later")
+
+	// 4. Enqueue tasks
+	_ = database.EnqueueTask("DISCOVER", map[string]interface{}{"target_id": target1.ID, "target_url": target1.TargetURL})
+	_ = database.EnqueueTask("DISCOVER", map[string]interface{}{"target_id": target2.ID, "target_url": target2.TargetURL})
+	_ = database.EnqueueTask("PROCESS", map[string]interface{}{"raw_job_id": job1.ID})
+	_ = database.EnqueueTask("PROCESS", map[string]interface{}{"raw_job_id": job3.ID})
+
+	// 5. Delete target 1
+	deletedTarget, deletedJobsCount, err := database.DeleteTarget(target1.ID)
+	if err != nil {
+		t.Fatalf("DeleteTarget failed: %v", err)
+	}
+	if deletedTarget.ID != target1.ID {
+		t.Errorf("expected deleted target ID %d, got %d", target1.ID, deletedTarget.ID)
+	}
+	if deletedJobsCount != 2 {
+		t.Errorf("expected 2 deleted jobs, got %d", deletedJobsCount)
+	}
+
+	// 6. Verify Target 1 is gone, Target 2 remains
+	allTargets, err := database.GetAllTargets()
+	if err != nil {
+		t.Fatalf("GetAllTargets failed: %v", err)
+	}
+	if len(allTargets) != 1 || allTargets[0].ID != target2.ID {
+		t.Errorf("expected 1 remaining target (target 2), got %v", allTargets)
+	}
+
+	// 7. Verify Target 1's jobs are gone, Target 2's job remains
+	jobsWithAnalysis, err := database.GetJobsWithAnalysis()
+	if err != nil {
+		t.Fatalf("GetJobsWithAnalysis failed: %v", err)
+	}
+	if len(jobsWithAnalysis) != 1 || jobsWithAnalysis[0].ID != job3.ID {
+		t.Errorf("expected only job 3 to remain, got %v", jobsWithAnalysis)
+	}
+
+	// 8. Verify applications for target 1 are gone, target 2 application remains
+	apps, err := database.GetApplications()
+	if err != nil {
+		t.Fatalf("GetApplications failed: %v", err)
+	}
+	if len(apps) != 1 || apps[0].ID != job3.ID {
+		t.Errorf("expected only job 3 application to remain, got %v", apps)
+	}
+}
+
+func TestRemoveTargetFromYAML(t *testing.T) {
+	tempDir := t.TempDir()
+	yamlPath := filepath.Join(tempDir, "targets.yaml")
+
+	initialYAML := `targets:
+  - name: "Google Seattle"
+    target_url: "https://google.com/jobs"
+    platform: "google_jobs"
+    is_active: true
+  - name: "Apple Seattle"
+    target_url: "https://apple.com/jobs"
+    platform: "apple"
+    is_active: true
+`
+	if err := os.WriteFile(yamlPath, []byte(initialYAML), 0644); err != nil {
+		t.Fatalf("failed to write initial yaml: %v", err)
+	}
+
+	// Remove Google Seattle
+	if err := RemoveTargetFromYAML(yamlPath, "https://google.com/jobs"); err != nil {
+		t.Fatalf("RemoveTargetFromYAML failed: %v", err)
+	}
+
+	// Read back and verify
+	data, err := os.ReadFile(yamlPath)
+	if err != nil {
+		t.Fatalf("failed to read yaml back: %v", err)
+	}
+
+	content := string(data)
+	if strings.Contains(content, "https://google.com/jobs") {
+		t.Errorf("expected google target to be removed, but found in file:\n%s", content)
+	}
+	if !strings.Contains(content, "https://apple.com/jobs") {
+		t.Errorf("expected apple target to remain, but not found in file:\n%s", content)
+	}
+}
+
+func TestTouchJobLastSeenAndMaintenance(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_maintenance.db")
+
+	database, err := InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	// Insert jobs
+	job1, _, err := database.InsertRawJob("canon-1", nil, "Job 1", "Google", "Seattle", "https://google.com/1", "Software Engineer Go")
+	if err != nil || job1.ID == 0 {
+		t.Fatalf("failed to insert job 1: %v", err)
+	}
+	job2, _, err := database.InsertRawJob("canon-2", nil, "Job 2", "Apple", "Seattle", "https://apple.com/2", "Software Engineer Swift")
+	if err != nil {
+		t.Fatalf("failed to insert job 2: %v", err)
+	}
+	job3, _, err := database.InsertRawJob("canon-3", nil, "Job 3", "Meta", "Seattle", "https://meta.com/3", "Software Engineer React")
+	if err != nil {
+		t.Fatalf("failed to insert job 3: %v", err)
+	}
+
+	// 1. Touch last seen for canon-1
+	if err := database.TouchJobLastSeen("canon-1"); err != nil {
+		t.Fatalf("TouchJobLastSeen failed: %v", err)
+	}
+
+	// Make job2 and job3 artificially stale (20 days and 40 days old)
+	_, err = database.Exec(`UPDATE raw_jobs SET last_seen_at = datetime('now', '-20 days') WHERE id = ?`, job2.ID)
+	if err != nil {
+		t.Fatalf("failed to update last_seen_at: %v", err)
+	}
+	_, err = database.Exec(`UPDATE raw_jobs SET last_seen_at = datetime('now', '-40 days') WHERE id = ?`, job3.ID)
+	if err != nil {
+		t.Fatalf("failed to update last_seen_at: %v", err)
+	}
+
+	// Protect job3 in active pipeline (APPLIED)
+	_, _ = database.CreateOrUpdateApplication(job3.ID, "APPLIED", "Applied to Meta")
+
+	// 2. Enqueue expire tasks (> 14 days) with pipeline protection
+	queuedExpireCount, err := database.EnqueueExpireTasks(14, true)
+	if err != nil {
+		t.Fatalf("EnqueueExpireTasks failed: %v", err)
+	}
+	// job2 is stale and unprotected -> should be queued
+	// job3 is stale but protected (APPLIED) -> should NOT be queued
+	// job1 is fresh -> should NOT be queued
+	if queuedExpireCount != 1 {
+		t.Errorf("expected 1 job queued for expire, got %d", queuedExpireCount)
+	}
+
+	// 3. Directly expire job2
+	if err := database.ExpireJob(job2.ID, "Test expiration"); err != nil {
+		t.Fatalf("ExpireJob failed: %v", err)
+	}
+
+	// Verify job2 status in DB
+	j2Detail, err := database.GetJobByID(job2.ID)
+	if err != nil {
+		t.Fatalf("GetJobByID failed: %v", err)
+	}
+	if j2Detail.App == nil || j2Detail.App.Status != "EXPIRED" {
+		t.Errorf("expected job2 to have EXPIRED application, got %v", j2Detail.App)
+	}
+
+	// Verify job3 still APPLIED
+	j3Detail, err := database.GetJobByID(job3.ID)
+	if err != nil {
+		t.Fatalf("GetJobByID failed: %v", err)
+	}
+	if j3Detail.App == nil || j3Detail.App.Status != "APPLIED" {
+		t.Errorf("expected job3 to remain APPLIED, got %v", j3Detail.App)
+	}
+
+	// 4. Stats check
+	stats, err := database.GetJobMaintenanceStats()
+	if err != nil {
+		t.Fatalf("GetJobMaintenanceStats failed: %v", err)
+	}
+	if stats["total_jobs"] != 3 {
+		t.Errorf("expected total_jobs=3, got %d", stats["total_jobs"])
+	}
+	if stats["active_7d"] != 1 {
+		t.Errorf("expected active_7d=1, got %d", stats["active_7d"])
+	}
+	if stats["stale_7d"] != 2 {
+		t.Errorf("expected stale_7d=2, got %d", stats["stale_7d"])
+	}
+	if stats["expired_count"] != 1 {
+		t.Errorf("expected expired_count=1, got %d", stats["expired_count"])
+	}
+}
+
+func TestDevDBMigration(t *testing.T) {
+	devDBPath := "../../data/db/dev/hireme.db"
+	if _, err := os.Stat(devDBPath); os.IsNotExist(err) {
+		return
+	}
+	database, err := InitDB(devDBPath)
+	if err != nil {
+		t.Fatalf("InitDB dev db failed: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	jobs, err := database.GetJobsWithAnalysis()
+	if err != nil {
+		t.Fatalf("GetJobsWithAnalysis on dev db failed: %v", err)
+	}
+	t.Logf("Successfully verified %d jobs from dev db", len(jobs))
+}
+
+

internal/db/targets_loader.go

@@ -16,7 +16,7 @@ type targetConfig struct {
 	} `yaml:"targets"`
 }
 
-// LoadTargetsFromYAML reads a YAML file and inserts targets into the database if it's empty
+// LoadTargetsFromYAML reads a YAML file and inserts/syncs targets with the database
 func (db *DB) LoadTargetsFromYAML(filepath string) error {
 	data, err := os.ReadFile(filepath)
 	if err != nil {
@@ -28,21 +28,71 @@ func (db *DB) LoadTargetsFromYAML(filepath string) error {
 		return fmt.Errorf("could not parse target config: %w", err)
 	}
 
-	// Check if we already have targets
-	targets, err := db.GetAllTargets()
+	// Get existing targets to prevent duplicates and sync is_active if needed
+	existingTargets, err := db.GetAllTargets()
 	if err != nil {
 		return err
 	}
-	if len(targets) > 0 {
-		// Do not overwrite existing targets
-		return nil
+
+	existingURLs := make(map[string]*SearchTarget)
+	for i := range existingTargets {
+		existingURLs[existingTargets[i].TargetURL] = &existingTargets[i]
 	}
 
 	for _, t := range config.Targets {
-		_, err := db.CreateTarget(t.Name, t.TargetURL, t.Platform)
+		existing, found := existingURLs[t.TargetURL]
+		if !found {
+			_, err := db.CreateTarget(t.Name, t.TargetURL, t.Platform, t.IsActive)
 			if err != nil {
 				return fmt.Errorf("failed to insert target %s: %w", t.Name, err)
 			}
+		} else {
+			// Sync is_active state with targets.yaml on startup
+			if existing.IsActive != t.IsActive {
+				_ = db.SetTargetActive(existing.ID, t.IsActive)
+			}
+		}
 	}
 	return nil
 }
+
+// RemoveTargetFromYAML removes a target from the target configuration YAML file if it exists
+func RemoveTargetFromYAML(filepath string, targetURL string) error {
+	if filepath == "" {
+		return nil
+	}
+	data, err := os.ReadFile(filepath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil
+		}
+		return fmt.Errorf("could not read target config: %w", err)
+	}
+
+	var config targetConfig
+	if err := yaml.Unmarshal(data, &config); err != nil {
+		return fmt.Errorf("could not parse target config: %w", err)
+	}
+
+	var newTargets []struct {
+		Name      string `yaml:"name"`
+		TargetURL string `yaml:"target_url"`
+		Platform  string `yaml:"platform"`
+		IsActive  bool   `yaml:"is_active"`
+	}
+
+	for _, t := range config.Targets {
+		if t.TargetURL != targetURL {
+			newTargets = append(newTargets, t)
+		}
+	}
+
+	config.Targets = newTargets
+	outData, err := yaml.Marshal(&config)
+	if err != nil {
+		return fmt.Errorf("could not marshal target config: %w", err)
+	}
+
+	return os.WriteFile(filepath, outData, 0644)
+}
+

internal/db/targets_loader_test.go

@@ -0,0 +1,59 @@
+package db
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestLoadTargetsFromYAMLRespectsIsActive(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_targets_active.db")
+	yamlPath := filepath.Join(tempDir, "targets.yaml")
+
+	yamlContent := `targets:
+  - name: "Google Seattle"
+    target_url: "https://google.com/jobs"
+    platform: "google_jobs"
+    is_active: false
+  - name: "Apple Seattle"
+    target_url: "https://apple.com/jobs"
+    platform: "apple"
+    is_active: true
+`
+	if err := os.WriteFile(yamlPath, []byte(yamlContent), 0644); err != nil {
+		t.Fatalf("failed to write yaml: %v", err)
+	}
+
+	database, err := InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	if err := database.LoadTargetsFromYAML(yamlPath); err != nil {
+		t.Fatalf("LoadTargetsFromYAML failed: %v", err)
+	}
+
+	targets, err := database.GetAllTargets()
+	if err != nil {
+		t.Fatalf("GetAllTargets failed: %v", err)
+	}
+
+	if len(targets) != 2 {
+		t.Fatalf("expected 2 targets, got %d", len(targets))
+	}
+
+	for _, target := range targets {
+		if target.TargetURL == "https://google.com/jobs" {
+			if target.IsActive != false {
+				t.Errorf("expected Google target to have is_active=false, got %v", target.IsActive)
+			}
+		}
+		if target.TargetURL == "https://apple.com/jobs" {
+			if target.IsActive != true {
+				t.Errorf("expected Apple target to have is_active=true, got %v", target.IsActive)
+			}
+		}
+	}
+}

internal/db/tasks.go

@@ -4,7 +4,8 @@ import (
 	"database/sql"
 	"encoding/json"
 	"fmt"
-	"log"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
 )
 
 // QueueStats holds task queue counters
@@ -96,7 +97,7 @@ func (db *DB) FailTask(id int64, errStr string) error {
 	var retryCount int
 	err := db.QueryRow(`SELECT retry_count FROM tasks WHERE id = ?`, id).Scan(&retryCount)
 	if err != nil {
-		log.Printf("[DB] FailTask: failed to get retry_count for task %d: %v", id, err)
+		logger.Log.Errorw("FailTask: failed to get retry_count", "component", "DB", "task_id", id, "error", err)
 		return err
 	}
 

internal/filter/filter.go

@@ -3,7 +3,7 @@ package filter
 import (
 	"context"
 
-	"github.com/hireme/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
 )
 
 type FilterResult struct {

internal/filter/filter_test.go

@@ -4,7 +4,7 @@ import (
 	"context"
 	"testing"
 
-	"github.com/hireme/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
 )
 
 func TestHardKeywordFilter(t *testing.T) {
@@ -30,6 +30,39 @@ func TestHardKeywordFilter(t *testing.T) {
 	}
 }
 
+func TestScoringFilterWordBoundaries(t *testing.T) {
+	ctx := context.Background()
+	sf := NewScoringFilter([]KeywordWeight{
+		{Keyword: "go", Weight: 20.0},
+		{Keyword: "ai", Weight: 15.0},
+		{Keyword: "c++", Weight: 10.0},
+	})
+
+	// "algorithm" and "email" should NOT trigger "go" or "ai"
+	res1, err := sf.Process(ctx, &db.RawJob{
+		Title:      "Algorithm Developer",
+		RawContent: "Send email to apply for algorithm development.",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if res1.Score != 0.0 {
+		t.Errorf("expected score 0 for false-positive substrings, got %f", res1.Score)
+	}
+
+	// Real "go", "ai", and "c++"
+	res2, err := sf.Process(ctx, &db.RawJob{
+		Title:      "AI Runtime Engineer",
+		RawContent: "Must know Go and C++ programming.",
+	})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if res2.Score != 45.0 {
+		t.Errorf("expected score 45.0 (20+15+10), got %f", res2.Score)
+	}
+}
+
 func TestFilterPipeline(t *testing.T) {
 	ctx := context.Background()
 	hardFilter := NewHardKeywordFilter([]string{"Unpaid", "No Sponsorship"})

internal/filter/hard_keyword.go

@@ -3,25 +3,43 @@ package filter
 import (
 	"context"
 	"fmt"
+	"regexp"
 	"strings"
 
-	"github.com/hireme/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
 )
 
+type compiledBlacklist struct {
+	keyword string
+	regex   *regexp.Regexp
+}
+
 type HardKeywordFilter struct {
 	Blacklist []string
+	compiled  []compiledBlacklist
 }
 
 func NewHardKeywordFilter(blacklist []string) *HardKeywordFilter {
-	if len(blacklist) == 0 {
-		blacklist = []string{
-			"no sponsorship",
-			"us citizen only",
-			"unpaid",
-			"volunteer",
+	compiled := make([]compiledBlacklist, 0, len(blacklist))
+	for _, kw := range blacklist {
+		term := strings.TrimSpace(kw)
+		if term != "" {
+			quoted := regexp.QuoteMeta(term)
+			quoted = regexp.MustCompile(`\\\s+`).ReplaceAllString(quoted, `\s+`)
+			pattern := `(?i)(?:^|[^a-zA-Z0-9_])` + quoted + `(?:[^a-zA-Z0-9_]|$)`
+			if re, err := regexp.Compile(pattern); err == nil {
+				compiled = append(compiled, compiledBlacklist{
+					keyword: kw,
+					regex:   re,
+				})
 			}
 		}
-	return &HardKeywordFilter{Blacklist: blacklist}
+	}
+
+	return &HardKeywordFilter{
+		Blacklist: blacklist,
+		compiled:  compiled,
+	}
 }
 
 func (f *HardKeywordFilter) Name() string {
@@ -29,15 +47,14 @@ func (f *HardKeywordFilter) Name() string {
 }
 
 func (f *HardKeywordFilter) Process(ctx context.Context, job *db.RawJob) (FilterResult, error) {
-	combined := strings.ToLower(job.Title + " " + job.RawContent)
+	combined := job.Title + " " + job.RawContent
 
-	for _, kw := range f.Blacklist {
-		term := strings.ToLower(strings.TrimSpace(kw))
-		if term != "" && strings.Contains(combined, term) {
+	for _, cb := range f.compiled {
+		if cb.regex.MatchString(combined) {
 			return FilterResult{
 				Keep:   false,
 				Score:  0,
-				Reason: fmt.Sprintf("Contains blacklisted keyword: %q", kw),
+				Reason: fmt.Sprintf("Contains blacklisted keyword: %q", cb.keyword),
 			}, nil
 		}
 	}

internal/filter/scoring.go

@@ -3,37 +3,62 @@ package filter
 import (
 	"context"
 	"fmt"
+	"regexp"
 	"strings"
 
-	"github.com/hireme/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
 )
 
 type KeywordWeight struct {
-	Keyword string
-	Weight  float64
+	Keyword string  `yaml:"keyword"`
+	Weight  float64 `yaml:"weight"`
+}
+
+type compiledWeight struct {
+	keyword string
+	weight  float64
+	regex   *regexp.Regexp
 }
 
 type ScoringFilter struct {
 	Weights  []KeywordWeight
+	compiled []compiledWeight
+}
+
+func compilePattern(term string) *regexp.Regexp {
+	clean := strings.TrimSpace(term)
+	if clean == "" {
+		return nil
+	}
+	// Quote meta characters, replace multiple whitespace with \s+
+	quoted := regexp.QuoteMeta(clean)
+	quoted = regexp.MustCompile(`\\\s+`).ReplaceAllString(quoted, `\s+`)
+	// Boundary regex that works for both alphanumeric and punctuation (e.g. c++)
+	pattern := `(?i)(?:^|[^a-zA-Z0-9_])` + quoted + `(?:[^a-zA-Z0-9_]|$)`
+	re, err := regexp.Compile(pattern)
+	if err != nil {
+		return nil
+	}
+	return re
 }
 
 func NewScoringFilter(weights []KeywordWeight) *ScoringFilter {
-	if len(weights) == 0 {
-		weights = []KeywordWeight{
-			{Keyword: "go", Weight: 25.0},
-			{Keyword: "golang", Weight: 25.0},
-			{Keyword: "backend", Weight: 15.0},
-			{Keyword: "distributed systems", Weight: 20.0},
-			{Keyword: "microservices", Weight: 15.0},
-			{Keyword: "sqlite", Weight: 10.0},
-			{Keyword: "postgres", Weight: 10.0},
-			{Keyword: "api", Weight: 10.0},
-			{Keyword: "full stack", Weight: 10.0},
-			{Keyword: "rest", Weight: 5.0},
-			{Keyword: "grpc", Weight: 15.0},
+	compiled := make([]compiledWeight, 0, len(weights))
+	for _, kw := range weights {
+		re := compilePattern(kw.Keyword)
+		if re != nil {
+			compiled = append(compiled, compiledWeight{
+				keyword: kw.Keyword,
+				weight:  kw.Weight,
+				regex:   re,
+			})
 		}
 	}
-	return &ScoringFilter{Weights: weights}
+
+	return &ScoringFilter{
+		Weights:  weights,
+		compiled: compiled,
+	}
 }
 
 func (s *ScoringFilter) Name() string {
@@ -41,18 +66,24 @@ func (s *ScoringFilter) Name() string {
 }
 
 func (f *ScoringFilter) Process(ctx context.Context, job *db.RawJob) (FilterResult, error) {
-	combinedLower := strings.ToLower(job.Title + " " + job.RawContent)
+	combined := job.Title + " " + job.RawContent
 	var totalScore float64
 	var matched []string
 
-	for _, kw := range f.Weights {
-		term := strings.ToLower(strings.TrimSpace(kw.Keyword))
-		if term != "" && strings.Contains(combinedLower, term) {
-			totalScore += kw.Weight
-			matched = append(matched, kw.Keyword)
+	for _, cw := range f.compiled {
+		if cw.regex.MatchString(combined) {
+			totalScore += cw.weight
+			matched = append(matched, cw.keyword)
 		}
 	}
 
+	if totalScore < 0 {
+		totalScore = 0
+	}
+	if totalScore > 100 {
+		totalScore = 100
+	}
+
 	return FilterResult{
 		Keep:   true,
 		Score:  totalScore,

internal/ui/handlers.go

@@ -2,13 +2,14 @@ package ui
 
 import (
 	"encoding/json"
+	"fmt"
 	"net/http"
 	"strconv"
 
-	"github.com/hireme/hireme/internal/config"
-	"github.com/hireme/hireme/internal/db"
-	"github.com/hireme/hireme/internal/logger"
-	"github.com/hireme/hireme/internal/worker"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/config"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/worker"
 )
 
 type Server struct {
@@ -30,6 +31,8 @@ func (s *Server) RegisterRoutes(mux *http.ServeMux) {
 	mux.HandleFunc("GET /targets", s.handleGetTargets)
 	mux.HandleFunc("POST /targets", s.handleCreateTarget)
 	mux.HandleFunc("POST /targets/{id}/toggle", s.handleToggleTarget)
+	mux.HandleFunc("POST /targets/{id}/delete", s.handleDeleteTarget)
+	mux.HandleFunc("DELETE /targets/{id}", s.handleDeleteTarget)
 	mux.HandleFunc("GET /jobs", s.handleGetJobs)
 	mux.HandleFunc("GET /jobs/{id}", s.handleGetJobDetail)
 	mux.HandleFunc("POST /jobs/{id}/analyze", s.handleAnalyzeJob)
@@ -42,6 +45,8 @@ func (s *Server) RegisterRoutes(mux *http.ServeMux) {
 	mux.HandleFunc("POST /api/tasks/reprocess", s.handleReprocessJobs)
 	mux.HandleFunc("POST /api/tasks/toggle", s.handleToggleTasks)
 	mux.HandleFunc("GET /api/status", s.handleGetStatus)
+	mux.HandleFunc("POST /api/maintenance/stale-expire", s.handleMaintenanceStaleExpire)
+	mux.HandleFunc("GET /api/maintenance/stats", s.handleMaintenanceStats)
 }
 
 func (s *Server) handleIndex(w http.ResponseWriter, r *http.Request) {
@@ -68,27 +73,39 @@ func (s *Server) handleIndex(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	qualifiedCount := 0
+	toReviewCount := 0
+	inPipelineCount := 0
+
 	for _, j := range jobs {
-		if j.Analysis != nil && j.Analysis.IsQualified {
-			qualifiedCount++
+		if j.App != nil && j.App.Status != "OPTED_OUT" {
+			inPipelineCount++
+		} else if j.App == nil || j.App.Status != "OPTED_OUT" {
+			if j.Analysis != nil && (j.Analysis.IsQualified || j.Analysis.Status == "PENDING") {
+				toReviewCount++
 			}
 		}
+	}
+
+	mStats, err := s.db.GetJobMaintenanceStats()
+	if err != nil {
+		logger.Log.Warnw("Failed to get maintenance stats", "component", "UI", "error", err)
+	}
 
 	data := PageData{
 		Targets:            targets,
 		Jobs:               jobs,
 		Applications:       apps,
+		MaintenanceStats:   mStats,
 		ActiveTab:          "triage",
 		LLMProvider:        s.cfg.LLMProvider,
-		QualifiedJobsCount: qualifiedCount,
+		ToReviewCount:      toReviewCount,
 		TotalIngestedCount: len(jobs),
-		ActiveApplications: len(apps),
+		InPipelineCount:    inPipelineCount,
 	}
 
 	w.Header().Set("Content-Type", "text/html; charset=utf-8")
 	if err := RenderPage(w, data); err != nil {
-		logger.Log.Errorf("[UI] RenderPage error: %v", err)
+		logger.Log.Errorw("RenderPage error", "component", "UI", "error", err)
 	}
 }
 
@@ -118,7 +135,9 @@ func (s *Server) handleCreateTarget(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	_, err := s.db.CreateTarget(name, targetURL, platform)
+	logger.Log.Infow("Triggered /targets (Create)", "component", "API", "name", name, "platform", platform)
+
+	_, err := s.db.CreateTarget(name, targetURL, platform, true)
 	if err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
 		return
@@ -135,6 +154,8 @@ func (s *Server) handleToggleTarget(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	logger.Log.Infow("Triggered /targets/{id}/toggle", "component", "API", "target_id", id)
+
 	if err := s.db.ToggleTargetStatus(id); err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
 		return
@@ -143,6 +164,35 @@ func (s *Server) handleToggleTarget(w http.ResponseWriter, r *http.Request) {
 	s.handleGetTargets(w, r)
 }
 
+func (s *Server) handleDeleteTarget(w http.ResponseWriter, r *http.Request) {
+	idStr := r.PathValue("id")
+	id, err := strconv.ParseInt(idStr, 10, 64)
+	if err != nil {
+		http.Error(w, "invalid target id", http.StatusBadRequest)
+		return
+	}
+
+	logger.Log.Infow("Triggered /targets/{id}/delete", "component", "API", "target_id", id)
+
+	deletedTarget, deletedJobsCount, err := s.db.DeleteTarget(id)
+	if err != nil {
+		logger.Log.Errorw("Failed to delete target", "component", "API", "target_id", id, "error", err)
+		http.Error(w, err.Error(), http.StatusInternalServerError)
+		return
+	}
+
+	logger.Log.Infow("Target deleted successfully", "component", "API", "target_id", id, "name", deletedTarget.Name, "deleted_jobs", deletedJobsCount)
+
+	// Also remove from targets.yaml if target_config is configured
+	if s.cfg != nil && s.cfg.TargetConfig != "" && deletedTarget != nil {
+		if err := db.RemoveTargetFromYAML(s.cfg.TargetConfig, deletedTarget.TargetURL); err != nil {
+			logger.Log.Warnw("Could not remove target from targets.yaml", "component", "API", "error", err)
+		}
+	}
+
+	s.handleGetTargets(w, r)
+}
+
 func (s *Server) handleGetJobs(w http.ResponseWriter, r *http.Request) {
 	jobs, err := s.db.GetJobsWithAnalysis()
 	if err != nil {
@@ -180,7 +230,20 @@ func (s *Server) handleAnalyzeJob(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	// Enqueue analyze task instead of synchronous processing
+	logger.Log.Infow("Triggered /jobs/{id}/analyze", "component", "API", "job_id", id)
+
+	job, err := s.db.GetJobByID(id)
+	if err == nil && job != nil {
+		filterScore := 0.0
+		score := 0.0
+		if job.Analysis != nil {
+			filterScore = job.Analysis.FilterScore
+			score = job.Analysis.Score
+		}
+		_ = s.db.CreateJobAnalysis(id, filterScore, score, true, "PENDING")
+	}
+
+	// Enqueue analyze task
 	payload := worker.AnalyzePayload{RawJobID: int64(id)}
 	err = s.db.EnqueueTask("ANALYZE", payload)
 	if err != nil {
@@ -189,7 +252,7 @@ func (s *Server) handleAnalyzeJob(w http.ResponseWriter, r *http.Request) {
 	}
 
 	w.WriteHeader(http.StatusAccepted)
-	_, _ = w.Write([]byte(`<span class="text-xs px-2 py-1 bg-yellow-100 text-yellow-800 rounded">Analysis Queued</span>`))
+	_, _ = w.Write([]byte(`<span class="text-xs px-2.5 py-1 bg-amber-500/20 text-amber-300 border border-amber-500/30 rounded-lg font-semibold">🔄 AI Queued</span>`))
 }
 
 func (s *Server) handleOptOut(w http.ResponseWriter, r *http.Request) {
@@ -200,6 +263,8 @@ func (s *Server) handleOptOut(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	logger.Log.Infow("Triggered /jobs/{id}/opt-out", "component", "API", "job_id", id)
+
 	_, err = s.db.CreateOrUpdateApplication(id, "OPTED_OUT", "")
 	if err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
@@ -243,6 +308,8 @@ func (s *Server) handleCreateApplication(w http.ResponseWriter, r *http.Request)
 		return
 	}
 
+	logger.Log.Infow("Triggered /applications (Create)", "component", "API", "raw_job_id", rawJobID, "status", status)
+
 	_, err = s.db.CreateOrUpdateApplication(rawJobID, status, "")
 	if err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
@@ -272,6 +339,8 @@ func (s *Server) handleUpdateAppStatus(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	logger.Log.Infow("Triggered /applications/{id}/status (Update)", "component", "API", "app_id", appID, "new_status", newStatus)
+
 	// Fetch rawJobID for appID
 	apps, err := s.db.GetApplications()
 	if err != nil {
@@ -304,7 +373,7 @@ func (s *Server) handleUpdateAppStatus(w http.ResponseWriter, r *http.Request) {
 }
 
 func (s *Server) handleDiscoverTasks(w http.ResponseWriter, r *http.Request) {
-	logger.Log.Info("[API] Triggered /api/tasks/discover")
+	logger.Log.Infow("Triggered /api/tasks/discover", "component", "API")
 	targets, err := s.db.GetAllTargets()
 	if err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
@@ -331,7 +400,7 @@ func (s *Server) handleDiscoverTasks(w http.ResponseWriter, r *http.Request) {
 }
 
 func (s *Server) handleCleanTasks(w http.ResponseWriter, r *http.Request) {
-	logger.Log.Info("[API] Triggered /api/tasks/clean")
+	logger.Log.Infow("Triggered /api/tasks/clean", "component", "API")
 	err := s.db.ClearTaskQueue()
 	if err != nil {
 		http.Error(w, err.Error(), http.StatusInternalServerError)
@@ -342,9 +411,9 @@ func (s *Server) handleCleanTasks(w http.ResponseWriter, r *http.Request) {
 }
 
 func (s *Server) handleReprocessJobs(w http.ResponseWriter, r *http.Request) {
-	logger.Log.Info("[API] Triggered /api/tasks/reprocess")
+	logger.Log.Infow("Triggered /api/tasks/reprocess", "component", "API")
 	if err := s.db.ReprocessAllJobs(); err != nil {
-		logger.Log.Errorf("[API] Failed to reprocess jobs: %v", err)
+		logger.Log.Errorw("Failed to reprocess jobs", "component", "API", "error", err)
 		http.Error(w, err.Error(), http.StatusInternalServerError)
 		return
 	}
@@ -353,7 +422,7 @@ func (s *Server) handleReprocessJobs(w http.ResponseWriter, r *http.Request) {
 }
 
 func (s *Server) handleToggleTasks(w http.ResponseWriter, r *http.Request) {
-	logger.Log.Info("[API] Triggered /api/tasks/toggle")
+	logger.Log.Infow("Triggered /api/tasks/toggle", "component", "API")
 	isRunning := s.pool.Toggle()
 	status := "paused"
 	if isRunning {
@@ -385,3 +454,39 @@ func (s *Server) handleGetStatus(w http.ResponseWriter, r *http.Request) {
 	w.Header().Set("Content-Type", "application/json")
 	_ = json.NewEncoder(w).Encode(resp)
 }
+
+func (s *Server) handleMaintenanceStaleExpire(w http.ResponseWriter, r *http.Request) {
+	if err := r.ParseForm(); err != nil {
+		http.Error(w, err.Error(), http.StatusBadRequest)
+		return
+	}
+	days, _ := strconv.Atoi(r.FormValue("days"))
+	if days <= 0 {
+		days = 14
+	}
+	protect := r.FormValue("protect") != "false"
+
+	count, err := s.db.EnqueueExpireTasks(days, protect)
+	if err != nil {
+		logger.Log.Errorw("Failed to enqueue expire tasks", "component", "UI", "error", err)
+		http.Error(w, err.Error(), http.StatusInternalServerError)
+		return
+	}
+
+	w.Header().Set("Content-Type", "application/json")
+	_ = json.NewEncoder(w).Encode(map[string]interface{}{
+		"status":  "success",
+		"count":   count,
+		"message": fmt.Sprintf("Enqueued %d expire task(s) for postings not seen in %d+ days.", count, days),
+	})
+}
+
+func (s *Server) handleMaintenanceStats(w http.ResponseWriter, r *http.Request) {
+	stats, err := s.db.GetJobMaintenanceStats()
+	if err != nil {
+		http.Error(w, err.Error(), http.StatusInternalServerError)
+		return
+	}
+	w.Header().Set("Content-Type", "application/json")
+	_ = json.NewEncoder(w).Encode(stats)
+}

internal/ui/handlers_test.go

@@ -0,0 +1,134 @@
+package ui
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/config"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
+)
+
+func TestHandleDeleteTarget(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_ui_delete_target.db")
+	yamlPath := filepath.Join(tempDir, "targets.yaml")
+
+	initialYAML := `targets:
+  - name: "Test Target"
+    target_url: "https://example.com/test-jobs"
+    platform: "google_jobs"
+    is_active: true
+`
+	if err := osWriteFile(yamlPath, []byte(initialYAML)); err != nil {
+		t.Fatalf("failed to write yaml: %v", err)
+	}
+
+	database, err := db.InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	target, err := database.CreateTarget("Test Target", "https://example.com/test-jobs", "google_jobs", true)
+	if err != nil {
+		t.Fatalf("failed to create target: %v", err)
+	}
+
+	_, _, err = database.InsertRawJob("canon-ui-test", &target.ID, "Software Engineer", "Google", "Seattle", "https://example.com/job/1", "Go backend")
+	if err != nil {
+		t.Fatalf("failed to insert raw job: %v", err)
+	}
+
+	cfg := &config.Config{
+		TargetConfig: yamlPath,
+	}
+
+	server := NewServer(cfg, database, nil)
+	mux := http.NewServeMux()
+	server.RegisterRoutes(mux)
+
+	// Send POST /targets/{id}/delete
+	req := httptest.NewRequest("POST", "/targets/1/delete", nil)
+	w := httptest.NewRecorder()
+
+	mux.ServeHTTP(w, req)
+
+	if w.Code != http.StatusOK {
+		t.Fatalf("expected status 200, got %d: %s", w.Code, w.Body.String())
+	}
+
+	// Verify target table in response does not contain the deleted target
+	body := w.Body.String()
+	if strings.Contains(body, "Test Target") {
+		t.Errorf("expected target list in response to not contain 'Test Target', got:\n%s", body)
+	}
+
+	// Verify DB is empty of targets and jobs
+	targets, err := database.GetAllTargets()
+	if err != nil || len(targets) != 0 {
+		t.Errorf("expected 0 targets in DB, got %v (err: %v)", targets, err)
+	}
+
+	jobs, err := database.GetJobsWithAnalysis()
+	if err != nil || len(jobs) != 0 {
+		t.Errorf("expected 0 jobs in DB, got %v (err: %v)", jobs, err)
+	}
+}
+
+func osWriteFile(path string, data []byte) error {
+	return os.WriteFile(path, data, 0644)
+}
+
+func TestMaintenanceHandlers(t *testing.T) {
+	tempDir := t.TempDir()
+	dbPath := filepath.Join(tempDir, "test_ui_maintenance.db")
+
+	database, err := db.InitDB(dbPath)
+	if err != nil {
+		t.Fatalf("failed to init db: %v", err)
+	}
+	defer func() { _ = database.Close() }()
+
+	// Insert job
+	job, _, err := database.InsertRawJob("canon-m-1", nil, "Software Engineer", "Google", "Seattle", "https://example.com/job/1", "Go backend")
+	if err != nil {
+		t.Fatalf("failed to insert raw job: %v", err)
+	}
+
+	// Make stale (25 days old)
+	_, _ = database.Exec(`UPDATE raw_jobs SET last_seen_at = datetime('now', '-25 days') WHERE id = ?`, job.ID)
+
+	cfg := &config.Config{}
+	server := NewServer(cfg, database, nil)
+	mux := http.NewServeMux()
+	server.RegisterRoutes(mux)
+
+	// 1. Test POST /api/maintenance/stale-expire
+	req := httptest.NewRequest("POST", "/api/maintenance/stale-expire", strings.NewReader("days=14&protect=true"))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	w := httptest.NewRecorder()
+	mux.ServeHTTP(w, req)
+
+	if w.Code != http.StatusOK {
+		t.Fatalf("expected status 200, got %d: %s", w.Code, w.Body.String())
+	}
+	if !strings.Contains(w.Body.String(), `"status":"success"`) || !strings.Contains(w.Body.String(), `"count":1`) {
+		t.Errorf("unexpected body: %s", w.Body.String())
+	}
+
+	// 2. Test GET /api/maintenance/stats
+	reqStats := httptest.NewRequest("GET", "/api/maintenance/stats", nil)
+	wStats := httptest.NewRecorder()
+	mux.ServeHTTP(wStats, reqStats)
+
+	if wStats.Code != http.StatusOK {
+		t.Fatalf("expected status 200, got %d: %s", wStats.Code, wStats.Body.String())
+	}
+	if !strings.Contains(wStats.Body.String(), `"total_jobs":1`) {
+		t.Errorf("expected stats to show total_jobs:1, got %s", wStats.Body.String())
+	}
+}

internal/ui/template_apply.go

@@ -19,6 +19,8 @@ const applyTabHTML = `{{ define "applyTab" }}
                 <option value="INTERVIEW">💬 Interview</option>
                 <option value="OFFER">🎉 Offer</option>
                 <option value="REJECTED">❌ Rejected</option>
+                <option value="EXPIRED">⚠️ Expired / Closed</option>
+                <option value="OPTED_OUT">🚫 Opted Out</option>
                 <option value="all">All</option>
             </select>
             <button hx-get="/applications" hx-target="#apply-container" hx-swap="innerHTML"
@@ -69,6 +71,15 @@ const applyTableHTML = `{{ define "applyTable" }}
 
                 <!-- Pipeline Stage (arrow indicator) -->
                 <td class="px-4 py-3">
+                    {{ if eq .App.Status "EXPIRED" }}
+                    <span class="px-2.5 py-0.5 rounded-full bg-amber-500/15 border border-amber-500/30 text-amber-300 text-[10px] font-bold">
+                        ⚠️ Expired / Closed
+                    </span>
+                    {{ else if eq .App.Status "OPTED_OUT" }}
+                    <span class="px-2.5 py-0.5 rounded-full bg-slate-800 border border-slate-700 text-slate-400 text-[10px] font-medium">
+                        🚫 Opted Out
+                    </span>
+                    {{ else }}
                     <div class="flex items-center gap-0.5 text-[10px] font-semibold">
                         <span class="px-2 py-0.5 rounded-l-full {{ if eq .App.Status "SAVED" }}bg-slate-500 text-white{{ else }}bg-slate-800/60 text-slate-600{{ end }}">Saved</span>
                         <span class="text-slate-700">▶</span>
@@ -84,13 +95,14 @@ const applyTableHTML = `{{ define "applyTable" }}
                         <span class="px-2 py-0.5 rounded-r-full bg-slate-800/60 text-slate-600">Offer</span>
                         {{ end }}
                     </div>
+                    {{ end }}
                 </td>
 
                 <!-- Score -->
                 <td class="px-4 py-3" data-sort-val="{{ if .Analysis }}{{ printf "%.0f" .Analysis.Score }}{{ else }}-1{{ end }}">
                     {{ if .Analysis }}
-                    <span class="font-bold {{ if .Analysis.IsQualified }}text-emerald-400{{ else }}text-slate-500{{ end }}">
-                        {{ printf "%.0f" .Analysis.Score }}
+                    <span class="font-bold {{ if .Analysis.IsQualified }}text-emerald-400{{ else }}text-slate-400{{ end }}">
+                        {{ if eq .Analysis.Status "PARSED" }}🤖 {{ printf "%.0f" .Analysis.Score }}{{ else }}⚡ {{ printf "%.0f" .Analysis.FilterScore }}{{ end }}
                     </span>
                     {{ else }}
                     <span class="text-slate-600">–</span>
@@ -113,6 +125,7 @@ const applyTableHTML = `{{ define "applyTable" }}
                 <!-- Change Status -->
                 <td class="px-4 py-3 text-right">
                     <select hx-patch="/applications/{{ .App.ID }}/status"
+                            name="status"
                             hx-target="#apply-container"
                             hx-swap="innerHTML"
                             class="px-2 py-1 bg-slate-900 border border-slate-700 rounded-lg text-slate-200 text-[10px] focus:outline-none focus:border-indigo-500">
@@ -121,6 +134,8 @@ const applyTableHTML = `{{ define "applyTable" }}
                         <option value="INTERVIEW" {{ if eq .App.Status "INTERVIEW" }}selected{{ end }}>💬 Interview</option>
                         <option value="OFFER"     {{ if eq .App.Status "OFFER"     }}selected{{ end }}>🎉 Offer</option>
                         <option value="REJECTED"  {{ if eq .App.Status "REJECTED"  }}selected{{ end }}>❌ Rejected</option>
+                        <option value="EXPIRED"   {{ if eq .App.Status "EXPIRED"   }}selected{{ end }}>⚠️ Expired</option>
+                        <option value="OPTED_OUT" {{ if eq .App.Status "OPTED_OUT" }}selected{{ end }}>🚫 Opted Out</option>
                     </select>
                 </td>
             </tr>

internal/ui/template_jobs.go

@@ -18,6 +18,7 @@ const jobsTabHTML = `{{ define "jobsTab" }}
                 <option value="not-qualified">❌ Not Qualified</option>
                 <option value="saved">💾 Saved / In Pipeline</option>
                 <option value="all">🌐 All Jobs</option>
+                <option value="expired">⚠️ Expired / Closed</option>
                 <option value="opted-out">🚫 Opted-out</option>
             </select>
             <button hx-get="/jobs" hx-target="#jobs-container" hx-swap="innerHTML"
@@ -42,10 +43,10 @@ const jobListHTML = `{{ define "jobs" }}
             <tr>
                 <th class="px-4 py-3" data-sort="title">Job Title &amp; Company</th>
                 <th class="px-4 py-3 text-center">Link</th>
-                <th class="px-4 py-3" data-sort="status">Analysis</th>
+                <th class="px-4 py-3" data-sort="status">Status</th>
                 <th class="px-4 py-3" data-sort="score">Score &amp; Qualification</th>
                 <th class="px-4 py-3">Reason / Tags</th>
-                <th class="px-4 py-3" data-sort="date">Crawled</th>
+                <th class="px-4 py-3" data-sort="date">Last Updated</th>
                 <th class="px-4 py-3 text-right">Actions</th>
             </tr>
         </thead>
@@ -54,7 +55,7 @@ const jobListHTML = `{{ define "jobs" }}
             {{- $qual := "pending" -}}
             {{- if .Analysis }}{{ if .Analysis.IsQualified }}{{ $qual = "qualified" }}{{ else }}{{ $qual = "not-qualified" }}{{ end }}{{ end -}}
             {{- $appClass := "none" -}}
-            {{- if .App }}{{ if eq .App.Status "OPTED_OUT" }}{{ $appClass = "opted-out" }}{{ else }}{{ $appClass = "pipeline" }}{{ end }}{{ end -}}
+            {{- if .App }}{{ if eq .App.Status "OPTED_OUT" }}{{ $appClass = "opted-out" }}{{ else if eq .App.Status "EXPIRED" }}{{ $appClass = "expired" }}{{ else }}{{ $appClass = "pipeline" }}{{ end }}{{ end -}}
             <tr class="hover:bg-slate-800/40 transition-colors group" data-row data-qual="{{ $qual }}" data-app="{{ $appClass }}">
 
                 <!-- Title & Company -->
@@ -85,9 +86,13 @@ const jobListHTML = `{{ define "jobs" }}
                 <td class="px-4 py-3" data-sort-val="{{ if .Analysis }}{{ .Analysis.Status }}{{ else }}UNANALYZED{{ end }}">
                     {{ if .Analysis }}
                         {{ if eq .Analysis.Status "PARSED" }}
-                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-indigo-500/20 text-indigo-300 border border-indigo-500/30">PARSED</span>
+                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-indigo-500/20 text-indigo-300 border border-indigo-500/30">AI PARSED</span>
                         {{ else if eq .Analysis.Status "PENDING" }}
-                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-amber-500/20 text-amber-400 border border-amber-500/30">PENDING</span>
+                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-amber-500/20 text-amber-400 border border-amber-500/30">QUEUED</span>
+                        {{ else if eq .Analysis.Status "FILTERED_OUT" }}
+                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-slate-700/40 text-slate-400 border border-slate-600/30">LOW FIT</span>
+                        {{ else if eq .Analysis.Status "REJECTED" }}
+                        <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-rose-500/20 text-rose-400 border border-rose-500/30">REJECTED</span>
                         {{ else if eq .Analysis.Status "DONE" }}
                         <span class="inline-flex items-center px-2 py-0.5 rounded text-[10px] font-bold bg-emerald-500/20 text-emerald-400 border border-emerald-500/30">DONE</span>
                         {{ else }}
@@ -101,13 +106,31 @@ const jobListHTML = `{{ define "jobs" }}
                 <!-- Score & Qualification -->
                 <td class="px-4 py-3" data-sort-val="{{ if .Analysis }}{{ printf "%.0f" .Analysis.Score }}{{ else }}-1{{ end }}">
                     {{ if .Analysis }}
+                        {{ if eq .Analysis.Status "PARSED" }}
                             {{ if .Analysis.IsQualified }}
-                        <span class="inline-flex items-center px-2.5 py-1 rounded-full text-xs font-black border {{ scoreBadgeClass .Analysis.Score }}">
-                            ★ {{ printf "%.0f" .Analysis.Score }}
+                            <span class="inline-flex items-center px-2.5 py-1 rounded-full text-xs font-black border {{ scoreBadgeClass .Analysis.Score }}" title="AI Match Score (DeepSeek)">
+                                🤖 {{ printf "%.0f" .Analysis.Score }}
                             </span>
                             {{ else }}
+                            <span class="inline-flex items-center px-2 py-0.5 rounded-full text-[10px] font-semibold bg-rose-500/10 text-rose-400 border border-rose-500/20" title="AI Score: {{ printf "%.0f" .Analysis.Score }}">
+                                ✗ {{ printf "%.0f" .Analysis.Score }} Not Qual
+                            </span>
+                            {{ end }}
+                        {{ else if eq .Analysis.Status "PENDING" }}
+                            <span class="inline-flex items-center px-2.5 py-1 rounded-full text-xs font-semibold bg-amber-500/10 text-amber-300 border border-amber-500/30" title="Keyword Fit: {{ printf "%.0f" .Analysis.FilterScore }} - Waiting for AI">
+                                ⚡ {{ printf "%.0f" .Analysis.FilterScore }}
+                            </span>
+                        {{ else if eq .Analysis.Status "FILTERED_OUT" }}
+                            <span class="inline-flex items-center px-2 py-0.5 rounded-full text-[10px] font-medium bg-slate-800 text-slate-400 border border-slate-700" title="Keyword Fit: {{ printf "%.0f" .Analysis.FilterScore }}">
+                                ⚡ {{ printf "%.0f" .Analysis.FilterScore }}
+                            </span>
+                        {{ else if eq .Analysis.Status "REJECTED" }}
                             <span class="inline-flex items-center px-2 py-0.5 rounded-full text-[10px] font-semibold bg-rose-500/10 text-rose-400 border border-rose-500/20">
-                            ✗ Not Qualified
+                                🚫 Disqualified
+                            </span>
+                        {{ else }}
+                            <span class="inline-flex items-center px-2 py-0.5 rounded-full text-[10px] font-semibold bg-slate-700/50 text-slate-400 border border-slate-600/30">
+                                {{ printf "%.0f" .Analysis.Score }}
                             </span>
                         {{ end }}
                     {{ else }}
@@ -117,9 +140,13 @@ const jobListHTML = `{{ define "jobs" }}
 
                 <!-- Reason / Matched Tags -->
                 <td class="px-4 py-3 max-w-[180px]">
-                    {{ if .App }}{{ if eq .App.Status "OPTED_OUT" }}
-                    <span class="text-[10px] text-slate-500 italic">Opted out</span>
-                    {{ end }}{{ end }}
+                    {{ if .App }}
+                        {{ if eq .App.Status "OPTED_OUT" }}
+                        <span class="text-[10px] text-slate-500 italic block">Opted out</span>
+                        {{ else if eq .App.Status "EXPIRED" }}
+                        <span class="text-[10px] text-amber-400/80 italic block">⚠️ No longer active</span>
+                        {{ end }}
+                    {{ end }}
                     {{ if .Analysis }}
                         {{ if .Analysis.IsQualified }}
                             <div class="flex flex-wrap gap-1">
@@ -133,14 +160,16 @@ const jobListHTML = `{{ define "jobs" }}
                             {{ $missing := parseJSON .Analysis.MissingHighlights }}
                             {{ if $missing }}
                             <span class="text-[10px] text-rose-400/90 leading-snug line-clamp-2" title="{{ index $missing 0 }}">{{ index $missing 0 }}</span>
+                            {{ else if .Analysis.Summary }}
+                            <span class="text-[10px] text-slate-400/90 leading-snug line-clamp-2" title="{{ .Analysis.Summary }}">{{ .Analysis.Summary }}</span>
                             {{ end }}
                         {{ end }}
                     {{ end }}
                 </td>
 
                 <!-- Date -->
-                <td class="px-4 py-3 text-slate-400 text-[11px]" data-sort-val="{{ .CrawledAt.Unix }}">
-                    {{ formatDate .CrawledAt }}
+                <td class="px-4 py-3 text-slate-400 text-[11px]" data-sort-val="{{ .LastSeenAt.Unix }}">
+                    {{ formatDate .LastSeenAt }}
                 </td>
 
                 <!-- Actions: Save + Opt-out OR status badge -->
@@ -148,14 +177,32 @@ const jobListHTML = `{{ define "jobs" }}
                     {{ if .App }}
                         {{ if eq .App.Status "OPTED_OUT" }}
                         <!-- Opted out: show undo button -->
+                        <div class="flex items-center justify-end gap-1.5">
+                            <span class="text-[10px] text-slate-500">Opted-out</span>
                             <button hx-post="/applications"
                                     hx-vals='{"raw_job_id": {{ .ID }}, "status": "SAVED"}'
                                     hx-target="#jobs-container"
                                     hx-swap="innerHTML"
                                     title="Undo opt-out"
-                                class="px-2.5 py-1 rounded bg-slate-800 hover:bg-slate-700 text-slate-400 border border-slate-700 text-[10px] transition-colors">
-                            ↩ Undo
+                                    class="px-2 py-0.5 rounded bg-slate-800 hover:bg-slate-700 text-slate-300 border border-slate-700 text-[10px] transition-colors">
+                                ↩
                             </button>
+                        </div>
+                        {{ else if eq .App.Status "EXPIRED" }}
+                        <!-- Expired -->
+                        <div class="flex items-center justify-end gap-1.5">
+                            <span class="px-2 py-0.5 rounded bg-slate-800 text-slate-400 text-[10px] font-bold border border-slate-700">
+                                ⚠️ EXPIRED
+                            </span>
+                            <button hx-post="/applications"
+                                    hx-vals='{"raw_job_id": {{ .ID }}, "status": "SAVED"}'
+                                    hx-target="#jobs-container"
+                                    hx-swap="innerHTML"
+                                    title="Re-open / Save"
+                                    class="px-2 py-0.5 rounded bg-slate-800 hover:bg-slate-700 text-slate-300 border border-slate-700 text-[10px] transition-colors">
+                                ↩
+                            </button>
+                        </div>
                         {{ else }}
                         <!-- In pipeline -->
                         <span class="px-2 py-1 rounded bg-violet-500/20 text-violet-300 text-[10px] font-bold border border-violet-500/30">
@@ -203,29 +250,56 @@ const jobDetailHTML = `{{ define "jobDetail" }}
         <div class="flex-1 min-w-0 pr-4">
             <span class="text-xs font-bold text-indigo-400 uppercase tracking-wide">{{ .Company }}</span>
             <h2 class="text-xl font-extrabold text-white mt-0.5 leading-tight">{{ .Title }}</h2>
-            <p class="text-xs text-slate-400 mt-1">📍 {{ .Location }} &bull; Ingested {{ formatDate .CrawledAt }}</p>
+            <p class="text-xs text-slate-400 mt-1">📍 {{ .Location }} &bull; Last Updated {{ formatDate .LastSeenAt }}</p>
         </div>
         <button onclick="closeDetailModal()" class="flex-shrink-0 text-slate-400 hover:text-white text-lg p-1 rounded hover:bg-slate-800 transition-colors">✕</button>
     </div>
 
     {{ if .Analysis }}
-    <!-- Score + Status Grid -->
+    <!-- Dual Score Breakdown Grid -->
     <div class="grid grid-cols-2 gap-3">
+        <!-- AI Match Score Card -->
         <div class="p-4 rounded-xl bg-slate-800/60 border border-slate-700/60">
-            <span class="text-[10px] text-slate-400 uppercase tracking-wider font-semibold block">Match Score</span>
+            <div class="flex items-center justify-between">
+                <span class="text-[10px] text-indigo-300 uppercase tracking-wider font-bold">🤖 AI Match Score</span>
+                <span class="text-[10px] font-semibold px-2 py-0.5 rounded {{ if eq .Analysis.Status "PARSED" }}bg-indigo-500/20 text-indigo-300 border border-indigo-500/30{{ else }}bg-amber-500/20 text-amber-300 border border-amber-500/30{{ end }}">
+                    {{ .Analysis.Status }}
+                </span>
+            </div>
+            {{ if eq .Analysis.Status "PARSED" }}
                 {{ if .Analysis.IsQualified }}
-            <span class="text-2xl font-black text-emerald-400 mt-1 block">{{ printf "%.1f" .Analysis.Score }} <span class="text-sm font-medium text-slate-400">/ 100</span></span>
+                <span class="text-2xl font-black text-emerald-400 mt-1.5 block">{{ printf "%.1f" .Analysis.Score }} <span class="text-xs font-normal text-slate-400">/ 100</span></span>
+                <span class="text-[10px] text-emerald-400/90 font-medium">✓ Qualified for interview profile</span>
                 {{ else }}
-            <span class="text-base font-bold text-rose-400 mt-1 block">Not Qualified</span>
+                <span class="text-2xl font-black text-rose-400 mt-1.5 block">{{ printf "%.1f" .Analysis.Score }} <span class="text-xs font-normal text-slate-400">/ 100</span></span>
+                <span class="text-[10px] text-rose-400 font-medium">✗ Does not meet core fit threshold</span>
+                {{ end }}
+            {{ else }}
+                <span class="text-lg font-bold text-amber-300 mt-1.5 block">Pending Evaluation</span>
+                <span class="text-[10px] text-slate-400">DeepSeek LLM analysis queued</span>
             {{ end }}
         </div>
+
+        <!-- Keyword Relevance Card -->
         <div class="p-4 rounded-xl bg-slate-800/60 border border-slate-700/60">
-            <span class="text-[10px] text-slate-400 uppercase tracking-wider font-semibold block">Analysis Status</span>
-            <span class="text-sm font-bold text-indigo-300 mt-1 block uppercase">{{ .Analysis.Status }}</span>
+            <div class="flex items-center justify-between">
+                <span class="text-[10px] text-amber-400 uppercase tracking-wider font-bold">⚡ Keyword Fit</span>
+                <span class="text-[10px] text-slate-400">Pre-filter rule</span>
+            </div>
+            <span class="text-2xl font-black text-white mt-1.5 block">{{ printf "%.1f" .Analysis.FilterScore }} <span class="text-xs font-normal text-slate-400">/ 100</span></span>
+            <span class="text-[10px] text-slate-400">Lexical match against filters.yaml</span>
         </div>
     </div>
 
-    <!-- Highlights -->
+    <!-- AI Executive Summary -->
+    {{ if .Analysis.Summary }}
+    <div class="p-3.5 rounded-xl bg-indigo-950/40 border border-indigo-500/30">
+        <span class="text-[10px] font-bold text-indigo-300 uppercase tracking-wider block mb-1">💬 AI Executive Assessment</span>
+        <p class="text-xs text-indigo-100/90 leading-relaxed">{{ .Analysis.Summary }}</p>
+    </div>
+    {{ end }}
+
+    <!-- Highlights & Gaps -->
     <div class="space-y-4">
         {{ if .Analysis.IsQualified }}
         <div>

internal/ui/template_kanban.go

@@ -63,7 +63,7 @@ const kanbanCardHTML = `{{ define "kanbanCard" }}
     </div>
     {{ if .Analysis }}
     <div class="text-[10px] text-slate-400">
-        Score: <strong class="{{ scoreBadgeClass .Analysis.Score }}">{{ printf "%.0f" .Analysis.Score }}</strong>
+        Score: <strong class="{{ scoreBadgeClass .Analysis.Score }}">{{ if eq .Analysis.Status "PARSED" }}🤖 {{ else }}⚡ {{ end }}{{ printf "%.0f" .Analysis.Score }}</strong>
     </div>
     {{ end }}
     <div class="flex items-center justify-between pt-2 border-t border-slate-700/50 text-[10px]">

internal/ui/template_layout.go

@@ -68,17 +68,17 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
             <div class="hidden md:flex items-center space-x-6 text-xs">
                 <div class="flex items-center space-x-1.5">
                     <span class="w-2 h-2 rounded-full bg-emerald-400"></span>
-                    <span class="text-slate-400">Qualified:</span>
-                    <span class="font-bold text-emerald-400">{{ .QualifiedJobsCount }}</span>
+                    <span class="text-slate-400">To Review:</span>
+                    <span class="font-bold text-emerald-400">{{ .ToReviewCount }}</span>
                 </div>
                 <div class="flex items-center space-x-1.5">
-                    <span class="text-slate-400">Ingested:</span>
+                    <span class="text-slate-400">In Pipeline:</span>
+                    <span class="font-bold text-violet-400">{{ .InPipelineCount }}</span>
+                </div>
+                <div class="flex items-center space-x-1.5">
+                    <span class="text-slate-400">Total Ingested:</span>
                     <span class="font-bold text-white">{{ .TotalIngestedCount }}</span>
                 </div>
-                <div class="flex items-center space-x-1.5">
-                    <span class="text-slate-400">Applied:</span>
-                    <span class="font-bold text-violet-400">{{ .ActiveApplications }}</span>
-                </div>
                 <div class="flex items-center space-x-1.5 px-2 py-1 rounded-md bg-slate-900/60 border border-slate-800">
                     <span class="w-1.5 h-1.5 rounded-full bg-emerald-400 animate-pulse"></span>
                     <span class="text-slate-400">LLM:</span>
@@ -102,7 +102,11 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
                 </button>
                 <button id="tab-btn-targets" onclick="switchTab('targets')"
                         class="tab-btn px-5 py-3 text-sm font-semibold border-b-2 transition-colors text-slate-400 border-transparent hover:text-slate-200">
-                    ⚙️ Targets
+                    🎯 Targets
+                </button>
+                <button id="tab-btn-preferences" onclick="switchTab('preferences')"
+                        class="tab-btn px-5 py-3 text-sm font-semibold border-b-2 transition-colors text-slate-400 border-transparent hover:text-slate-200">
+                    ⚙️ Preferences
                 </button>
             </div>
         </div>
@@ -126,6 +130,11 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
             {{ template "targetsTab" . }}
         </div>
 
+        <!-- Tab: Preferences -->
+        <div id="tab-content-preferences" class="hidden space-y-5">
+            {{ template "preferencesTab" . }}
+        </div>
+
     </main>
 
     <!-- Right Slide-Over Detail Panel -->
@@ -137,6 +146,52 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
         </div>
     </div>
 
+    <!-- Delete Target Confirmation Modal -->
+    <div id="delete-target-modal" class="fixed inset-0 z-50 hidden flex items-center justify-center p-4" aria-modal="true" role="dialog">
+        <div class="fixed inset-0 bg-slate-950/80 backdrop-blur-sm transition-opacity" onclick="closeDeleteTargetModal()"></div>
+        <div class="relative w-full max-w-md bg-slate-900 border border-slate-700/80 rounded-2xl shadow-2xl shadow-black/80 p-6 text-slate-100 transform transition-all overflow-hidden z-10">
+            <div class="absolute -top-24 -left-24 w-48 h-48 bg-rose-500/15 rounded-full blur-3xl pointer-events-none"></div>
+            <div class="flex items-start gap-4">
+                <div class="w-10 h-10 rounded-xl bg-rose-500/10 border border-rose-500/30 flex items-center justify-center flex-shrink-0 text-xl shadow-inner">
+                    ⚠️
+                </div>
+                <div class="flex-1 min-w-0">
+                    <h3 class="text-base font-bold text-white leading-tight">Delete Target</h3>
+                    <p class="text-xs text-slate-400 mt-1">This is a bulk operation that cannot be undone.</p>
+                </div>
+                <button type="button" onclick="closeDeleteTargetModal()" class="text-slate-400 hover:text-white text-lg p-1 rounded hover:bg-slate-800 transition-colors">✕</button>
+            </div>
+
+            <div class="mt-4 p-3 rounded-xl bg-slate-950/60 border border-slate-800">
+                <span class="text-[10px] uppercase font-bold text-slate-500 tracking-wider block">Target to remove:</span>
+                <span id="delete-target-name" class="text-sm font-semibold text-rose-300 block mt-0.5 break-words"></span>
+            </div>
+
+            <div class="mt-3 text-xs text-slate-300 space-y-1.5 leading-relaxed">
+                <p class="text-rose-400 font-medium">⚠️ What will be removed:</p>
+                <ul class="list-disc list-inside text-[11px] text-slate-400 space-y-1 pl-1">
+                    <li>Target configuration and crawling settings</li>
+                    <li><strong class="text-slate-200">All discovered job postings</strong> from this target</li>
+                    <li>All associated AI qualification analyses &amp; scores</li>
+                    <li>Any <strong class="text-slate-200">in-pipeline applications</strong> tracked for these jobs</li>
+                </ul>
+            </div>
+
+            <div class="mt-6 flex items-center justify-end gap-3">
+                <button type="button"
+                        onclick="closeDeleteTargetModal()"
+                        class="px-4 py-2 text-xs font-semibold text-slate-300 hover:text-white bg-slate-800 hover:bg-slate-700 rounded-lg border border-slate-700 transition-colors">
+                    Cancel
+                </button>
+                <button id="delete-target-confirm-btn"
+                        type="button"
+                        class="px-4 py-2 text-xs font-semibold text-white bg-rose-600 hover:bg-rose-500 rounded-lg shadow-lg shadow-rose-500/25 transition-all flex items-center gap-1.5 active:scale-95">
+                    <span>🗑️</span> Delete Target &amp; All Posts
+                </button>
+            </div>
+        </div>
+    </div>
+
     <!-- Fixed Bottom Status & Action Bar -->
     <div id="bottom-bar"
          class="fixed bottom-0 left-0 right-0 z-40 bg-slate-900/95 backdrop-blur-md border-t border-slate-700/60 shadow-2xl shadow-black/50">
@@ -184,19 +239,25 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
 
     <script>
         // ── Tab Routing ──────────────────────────────────────────────────────
-        const TABS = ['jobs', 'apply', 'targets'];
+        const TABS = ['jobs', 'apply', 'targets', 'preferences'];
 
         function switchTab(tab) {
             TABS.forEach(t => {
-                document.getElementById('tab-content-' + t).classList.add('hidden');
+                const content = document.getElementById('tab-content-' + t);
+                if (content) content.classList.add('hidden');
                 const btn = document.getElementById('tab-btn-' + t);
+                if (btn) {
                     btn.classList.remove('text-indigo-400', 'border-indigo-500');
                     btn.classList.add('text-slate-400', 'border-transparent');
+                }
             });
-            document.getElementById('tab-content-' + tab).classList.remove('hidden');
+            const activeContent = document.getElementById('tab-content-' + tab);
+            if (activeContent) activeContent.classList.remove('hidden');
             const activeBtn = document.getElementById('tab-btn-' + tab);
+            if (activeBtn) {
                 activeBtn.classList.remove('text-slate-400', 'border-transparent');
                 activeBtn.classList.add('text-indigo-400', 'border-indigo-500');
+            }
             history.replaceState(null, '', '#' + tab);
         }
 
@@ -214,6 +275,97 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
             document.getElementById('detail-modal').classList.add('hidden');
         }
 
+        // ── Delete Target Modal ──────────────────────────────────────────────
+        let currentDeleteTargetId = null;
+
+        function openDeleteTargetModal(btn) {
+            const id = btn.dataset.targetId;
+            const name = btn.dataset.targetName || 'this target';
+            currentDeleteTargetId = id;
+            document.getElementById('delete-target-name').textContent = name;
+            const confirmBtn = document.getElementById('delete-target-confirm-btn');
+            confirmBtn.onclick = () => confirmDeleteTarget(id);
+            document.getElementById('delete-target-modal').classList.remove('hidden');
+        }
+
+        function closeDeleteTargetModal() {
+            document.getElementById('delete-target-modal').classList.add('hidden');
+            currentDeleteTargetId = null;
+        }
+
+        function confirmDeleteTarget(id) {
+            const confirmBtn = document.getElementById('delete-target-confirm-btn');
+            confirmBtn.disabled = true;
+            confirmBtn.textContent = 'Deleting…';
+
+            fetch('/targets/' + id + '/delete', { method: 'POST' })
+                .then(res => {
+                    if (!res.ok) throw new Error('Failed to delete target');
+                    return res.text();
+                })
+                .then(html => {
+                    const targetContainer = document.getElementById('targets-container');
+                    if (targetContainer) targetContainer.innerHTML = html;
+                    closeDeleteTargetModal();
+                    // Refresh Jobs and Apply tables & stats since associated jobs were deleted
+                    const jobsContainer = document.getElementById('jobs-container');
+                    if (jobsContainer) htmx.trigger(jobsContainer, 'refresh');
+                    const applyContainer = document.getElementById('apply-container');
+                    if (applyContainer) htmx.trigger(applyContainer, 'refresh');
+                    refreshStatus();
+                })
+                .catch(err => {
+                    alert('Error deleting target: ' + err.message);
+                })
+                .finally(() => {
+                    confirmBtn.disabled = false;
+                    confirmBtn.innerHTML = '<span>🗑️</span> Delete Target &amp; All Posts';
+                });
+        }
+
+        // ── Maintenance Actions ──────────────────────────────────────────────
+        function runMaintenanceAction(url, daysInputId, protectCheckboxId, btn, originalText) {
+            const days = document.getElementById(daysInputId)?.value || '14';
+            const protect = document.getElementById(protectCheckboxId)?.checked ?? true;
+            const alertBox = document.getElementById('maintenance-alert');
+
+            btn.disabled = true;
+            btn.textContent = 'Processing…';
+
+            fetch(url, {
+                method: 'POST',
+                headers: { 'Content-Type': 'application/x-www-form-urlencoded' },
+                body: new URLSearchParams({ days: days, protect: protect ? 'true' : 'false' })
+            })
+            .then(res => res.json())
+            .then(data => {
+                if (alertBox) {
+                    alertBox.className = 'p-4 rounded-xl border text-xs font-semibold ' + 
+                        (data.status === 'success' ? 'bg-emerald-950/60 border-emerald-500/40 text-emerald-300' : 'bg-rose-950/60 border-rose-500/40 text-rose-300');
+                    alertBox.innerHTML = '<span>✓</span> ' + (data.message || 'Operation completed successfully.');
+                    alertBox.classList.remove('hidden');
+                    setTimeout(() => { alertBox.classList.add('hidden'); }, 8000);
+                }
+                // Refresh Jobs and Apply tables & stats
+                const jobsContainer = document.getElementById('jobs-container');
+                if (jobsContainer) htmx.trigger(jobsContainer, 'refresh');
+                const applyContainer = document.getElementById('apply-container');
+                if (applyContainer) htmx.trigger(applyContainer, 'refresh');
+                refreshStatus();
+            })
+            .catch(err => {
+                if (alertBox) {
+                    alertBox.className = 'p-4 rounded-xl border text-xs font-semibold bg-rose-950/60 border-rose-500/40 text-rose-300';
+                    alertBox.innerHTML = '<span>✕</span> Error: ' + err.message;
+                    alertBox.classList.remove('hidden');
+                }
+            })
+            .finally(() => {
+                btn.disabled = false;
+                btn.textContent = originalText;
+            });
+        }
+
         // ── Bottom Bar: Status Polling ────────────────────────────────────────
         function refreshStatus() {
             fetch('/api/status')
@@ -298,14 +450,15 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
             table.querySelectorAll('tr[data-row]').forEach(row => {
                 const rowText = row.textContent.toLowerCase();
                 const rowQual = row.dataset.qual || '';   // qualified | not-qualified | pending
-                const rowApp  = row.dataset.app  || 'none'; // none | opted-out | pipeline
+                const rowApp  = row.dataset.app  || 'none'; // none | opted-out | expired | pipeline
                 const textMatch = !text || rowText.includes(text);
                 let qualMatch;
                 switch (qual) {
-                    case 'default':     qualMatch = rowApp !== 'opted-out' && (rowQual === 'qualified' || rowQual === 'pending'); break;
-                    case 'not-qualified': qualMatch = rowApp !== 'opted-out' && rowQual === 'not-qualified'; break;
+                    case 'default':     qualMatch = rowApp !== 'opted-out' && rowApp !== 'expired' && rowApp !== 'pipeline' && (rowQual === 'qualified' || rowQual === 'pending'); break;
+                    case 'not-qualified': qualMatch = rowApp !== 'opted-out' && rowApp !== 'expired' && rowQual === 'not-qualified'; break;
                     case 'saved':       qualMatch = rowApp === 'pipeline'; break;
-                    case 'all':         qualMatch = rowApp !== 'opted-out'; break;
+                    case 'all':         qualMatch = true; break;
+                    case 'expired':     qualMatch = rowApp === 'expired'; break;
                     case 'opted-out':   qualMatch = rowApp === 'opted-out'; break;
                     default:            qualMatch = true;
                 }
@@ -369,6 +522,14 @@ const mainLayoutHTML = `{{ define "page" }}<!DOCTYPE html>
             if (e.target.id === 'apply-container' || e.target.closest?.('#apply-container'))
                 initApplyTable();
         });
+
+        // Dismiss modals on Escape key
+        document.addEventListener('keydown', (e) => {
+            if (e.key === 'Escape') {
+                closeDetailModal();
+                closeDeleteTargetModal();
+            }
+        });
     </script>
 </body>
 </html>{{ end }}`

internal/ui/template_preferences.go

@@ -0,0 +1,91 @@
+package ui
+
+const preferencesTabHTML = `{{ define "preferencesTab" }}
+<div class="space-y-6 max-w-4xl">
+    <div>
+        <h2 class="text-base font-bold text-white flex items-center gap-2">
+            <span>⚙️</span> System Setup &amp; Maintenance
+        </h2>
+        <p class="text-xs text-slate-400 mt-0.5">Monitor database health and purge stale job postings not seen on target career sites</p>
+    </div>
+
+    <!-- Maintenance Result Alert Banner -->
+    <div id="maintenance-alert" class="hidden p-4 rounded-xl border transition-all"></div>
+
+    <!-- Statistics Overview Grid -->
+    <div class="grid grid-cols-2 sm:grid-cols-3 lg:grid-cols-6 gap-3">
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-slate-400 block tracking-wider">Total Ingested</span>
+            <span class="text-xl font-extrabold text-white mt-1 block">{{ .MaintenanceStats.total_jobs }}</span>
+        </div>
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-emerald-400 block tracking-wider">Active (&le;7d)</span>
+            <span class="text-xl font-extrabold text-emerald-400 mt-1 block">{{ .MaintenanceStats.active_7d }}</span>
+        </div>
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-amber-400 block tracking-wider">Stale (&gt;7d)</span>
+            <span class="text-xl font-extrabold text-amber-400 mt-1 block">{{ .MaintenanceStats.stale_7d }}</span>
+        </div>
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-rose-400 block tracking-wider">Stale (&gt;30d)</span>
+            <span class="text-xl font-extrabold text-rose-400 mt-1 block">{{ .MaintenanceStats.stale_30d }}</span>
+        </div>
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-slate-400 block tracking-wider">Expired / Closed</span>
+            <span class="text-xl font-extrabold text-slate-300 mt-1 block">{{ .MaintenanceStats.expired_count }}</span>
+        </div>
+        <div class="p-3.5 rounded-xl bg-slate-900/70 border border-slate-800">
+            <span class="text-[10px] uppercase font-bold text-violet-400 block tracking-wider">In Pipeline</span>
+            <span class="text-xl font-extrabold text-violet-300 mt-1 block">{{ .MaintenanceStats.in_pipeline }}</span>
+        </div>
+    </div>
+
+    <!-- Maintenance Tool: Purge Stale Postings (Worker Queue Driven) -->
+    <div class="p-6 rounded-xl bg-slate-900/60 border border-slate-800 space-y-5">
+        <div class="flex items-start justify-between">
+            <div class="flex items-center gap-3">
+                <span class="w-9 h-9 rounded-xl bg-amber-500/10 border border-amber-500/30 flex items-center justify-center text-lg">⚠️</span>
+                <div>
+                    <h3 class="text-sm font-bold text-white">Purge Stale Postings (Mark Expired)</h3>
+                    <p class="text-xs text-slate-400 mt-0.5">Enqueues worker queue tasks to mark job postings as <strong class="text-amber-300 font-semibold">Expired</strong> if they have not been discovered on target career boards for X days.</p>
+                </div>
+            </div>
+        </div>
+
+        <div class="pt-4 border-t border-slate-800/80 grid grid-cols-1 sm:grid-cols-3 gap-4 items-end">
+            <div>
+                <label for="expire-days" class="block text-xs font-semibold text-slate-300 mb-1.5">Not updated in:</label>
+                <select id="expire-days" class="w-full px-3 py-2 text-xs rounded-lg bg-slate-800 border border-slate-700 text-white focus:outline-none focus:border-indigo-500">
+                    <option value="7">7+ days</option>
+                    <option value="14" selected>14+ days</option>
+                    <option value="30">30+ days</option>
+                    <option value="60">60+ days</option>
+                </select>
+            </div>
+
+            <div class="flex items-center gap-2 pb-2">
+                <input type="checkbox" id="expire-protect-pipeline" checked class="rounded bg-slate-800 border-slate-700 text-indigo-600 focus:ring-0">
+                <label for="expire-protect-pipeline" class="text-xs text-slate-300">Protect active applications</label>
+            </div>
+
+            <div>
+                <button type="button"
+                        onclick="runMaintenanceAction('/api/maintenance/stale-expire', 'expire-days', 'expire-protect-pipeline', this, '⚠️ Queue Expire Tasks')"
+                        class="w-full py-2 bg-amber-600 hover:bg-amber-500 text-xs font-semibold text-white rounded-lg shadow-md transition-all active:scale-95">
+                    ⚠️ Queue Expire Tasks
+                </button>
+            </div>
+        </div>
+    </div>
+
+    <!-- Explanation Card -->
+    <div class="p-5 rounded-xl bg-slate-900/40 border border-slate-800/80 text-xs text-slate-400 space-y-2">
+        <h4 class="font-bold text-slate-200 uppercase tracking-wider text-[11px]">💡 How Job Verification Works</h4>
+        <p>
+            Every time <strong class="text-indigo-300">🔍 Discover</strong> runs, Hireme searches live listings directly on career sites (Google, Apple, Meta).
+            When an existing job appears in search results, its <strong class="text-emerald-400">Last Updated</strong> timestamp is refreshed.
+            If a role closes or is removed from search results, its Last Updated timestamp remains unchanged, allowing you to safely purge expired postings using the tool above.
+        </p>
+    </div>
+</div>
+{{ end }}`

internal/ui/template_targets.go

@@ -22,7 +22,8 @@ const targetsTabHTML = `{{ define "targetsTab" }}
             <select name="platform"
                     class="px-3 py-2 text-xs rounded-lg bg-slate-800 border border-slate-700 text-white focus:outline-none focus:border-indigo-500">
                 <option value="google_jobs">🔍 Google Jobs</option>
-                <!-- More platforms will be added as crawlers are implemented -->
+                <option value="apple">🍏 Apple</option>
+                <option value="meta">♾️ Meta</option>
             </select>
             <button type="submit"
                     class="px-4 py-2 bg-indigo-600 hover:bg-indigo-500 text-xs font-semibold text-white rounded-lg shadow-md transition-all">
@@ -68,12 +69,21 @@ const targetListHTML = `{{ define "targets" }}
                     {{ end }}
                 </td>
                 <td class="px-4 py-3 text-right">
+                    <div class="flex items-center justify-end gap-1.5">
                         <button hx-post="/targets/{{ .ID }}/toggle"
                                 hx-target="#targets-container"
                                 hx-swap="innerHTML"
                                 class="px-2.5 py-1 text-[11px] rounded bg-slate-800 hover:bg-slate-700 text-slate-200 border border-slate-700 transition-colors">
                             {{ if .IsActive }}Pause{{ else }}Activate{{ end }}
                         </button>
+                        <button type="button"
+                                data-target-id="{{ .ID }}"
+                                data-target-name="{{ .Name }}"
+                                onclick="openDeleteTargetModal(this)"
+                                class="px-2.5 py-1 text-[11px] rounded bg-rose-500/10 hover:bg-rose-500/20 text-rose-400 border border-rose-500/30 hover:border-rose-500/50 transition-colors flex items-center gap-1">
+                            <span>🗑️</span> Delete
+                        </button>
+                    </div>
                 </td>
             </tr>
             {{ else }}

internal/ui/templates.go

@@ -8,7 +8,7 @@ import (
 	"strings"
 	"time"
 
-	"github.com/hireme/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
 )
 
 var funcMap = template.FuncMap{
@@ -43,6 +43,10 @@ var funcMap = template.FuncMap{
 			return "💼 LinkedIn"
 		} else if strings.Contains(p, "google") {
 			return "🔍 Google Jobs"
+		} else if strings.Contains(p, "apple") {
+			return "🍏 Apple"
+		} else if strings.Contains(p, "meta") || strings.Contains(p, "facebook") {
+			return "♾️ Meta"
 		} else if strings.Contains(p, "greenhouse") {
 			return "🌿 Greenhouse"
 		} else if strings.Contains(p, "lever") {
@@ -68,11 +72,12 @@ type PageData struct {
 	Targets            []db.SearchTarget
 	Jobs               []db.JobWithAnalysis
 	Applications       []db.JobWithAnalysis
+	MaintenanceStats   map[string]int
 	ActiveTab          string
 	LLMProvider        string
-	QualifiedJobsCount int
+	ToReviewCount      int
 	TotalIngestedCount int
-	ActiveApplications int
+	InPipelineCount    int
 }
 
 var rootTemplate *template.Template
@@ -88,6 +93,7 @@ func init() {
 		jobDetailHTML,
 		targetsTabHTML,
 		targetListHTML,
+		preferencesTabHTML,
 		applyTabHTML,
 		applyTableHTML,
 		kanbanBoardHTML,
@@ -127,3 +133,8 @@ func RenderKanbanBoard(w io.Writer, apps []db.JobWithAnalysis) error {
 func RenderApplyTable(w io.Writer, apps []db.JobWithAnalysis) error {
 	return rootTemplate.ExecuteTemplate(w, "applyTable", apps)
 }
+
+func RenderPreferencesTab(w io.Writer, data PageData) error {
+	return rootTemplate.ExecuteTemplate(w, "preferencesTab", data)
+}
+

internal/worker/pool.go

@@ -9,19 +9,32 @@ import (
 	"sync/atomic"
 	"time"
 
-	"github.com/hireme/hireme/internal/analyzer"
-	"github.com/hireme/hireme/internal/config"
-	"github.com/hireme/hireme/internal/crawler"
-	"github.com/hireme/hireme/internal/db"
-	"github.com/hireme/hireme/internal/filter"
-	"github.com/hireme/hireme/internal/logger"
-	"github.com/hireme/hireme/pkg/crawler/google"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/analyzer"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/config"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/db"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/filter"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/amazon"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/apple"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/bytedance"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/databricks"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/google"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/linkedin"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/meta"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/microsoft"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/netflix"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/nvidia"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/snowflake"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/stripe"
+	"git.pengzhan.dev/haopengzhan/hireme/pkg/crawler/uber"
 )
 
 type WorkerPool struct {
 	db                *db.DB
 	analyzer          analyzer.LLMAnalyzer
 	filterPipe        *filter.FilterPipeline
+	analysisThreshold float64
 	isRunning         atomic.Bool
 	workerCount       int
 	wg                sync.WaitGroup
@@ -29,17 +42,24 @@ type WorkerPool struct {
 	cancel            context.CancelFunc
 }
 
-func NewWorkerPool(database *db.DB, analyzer analyzer.LLMAnalyzer, numWorkers int, filterCfg *config.FilterConfig) *WorkerPool {
-	// Initialize filters using the provided config
+func NewWorkerPool(database *db.DB, analyzer analyzer.LLMAnalyzer, numWorkers int, threshold float64, filterCfg *config.FilterConfig) *WorkerPool {
 	filterPipe := filter.NewPipeline(
 		filter.NewHardKeywordFilter(filterCfg.HardKeywords),
 		filter.NewScoringFilter(filterCfg.ScoringWeights),
 	)
 
+	if threshold <= 0 {
+		threshold = filterCfg.AnalysisThreshold
+	}
+	if threshold <= 0 {
+		threshold = 25.0
+	}
+
 	return &WorkerPool{
 		db:                database,
 		analyzer:          analyzer,
 		filterPipe:        filterPipe,
+		analysisThreshold: threshold,
 		workerCount:       numWorkers,
 	}
 }
@@ -52,48 +72,52 @@ func (p *WorkerPool) Toggle() bool {
 	if p.IsRunning() {
 		p.Stop()
 		return false
-	}
+	} else {
 		p.Start()
 		return true
 	}
+}
 
 func (p *WorkerPool) Start() {
-	if p.isRunning.Load() {
-		return
+	if p.isRunning.Swap(true) {
+		return // Already running
 	}
+
 	p.ctx, p.cancel = context.WithCancel(context.Background())
-	p.isRunning.Store(true)
+	logger.Log.Infow("Started workers", "component", "WorkerPool", "count", p.workerCount)
 
 	for i := 0; i < p.workerCount; i++ {
 		p.wg.Add(1)
 		go p.workerLoop(i)
 	}
-	logger.Log.Infof("[WorkerPool] Started %d workers", p.workerCount)
 }
 
 func (p *WorkerPool) Stop() {
-	if !p.isRunning.Load() {
-		return
+	if !p.isRunning.Swap(false) {
+		return // Already stopped
 	}
+
+	logger.Log.Infow("Stopping worker pool...", "component", "WorkerPool")
+	if p.cancel != nil {
 		p.cancel()
-	p.isRunning.Store(false)
+	}
 	p.wg.Wait()
-	logger.Log.Infof("[WorkerPool] All workers stopped")
+	logger.Log.Infow("Worker pool stopped", "component", "WorkerPool")
 }
 
 func (p *WorkerPool) workerLoop(workerID int) {
 	defer p.wg.Done()
-	logger.Log.Debugf("[Worker %d] Starting loop", workerID)
+	logger.Log.Debugw("Starting loop", "component", "Worker", "worker_id", workerID)
 
 	for {
 		select {
 		case <-p.ctx.Done():
-			logger.Log.Debugf("[Worker %d] Context cancelled, exiting", workerID)
+			logger.Log.Debugw("Context cancelled, exiting", "component", "Worker", "worker_id", workerID)
 			return
 		default:
 			task, err := p.db.ClaimNextTask()
 			if err != nil {
-				logger.Log.Errorf("[Worker %d] Error claiming task: %v", workerID, err)
+				logger.Log.Errorw("Failed to claim task", "component", "Worker", "worker_id", workerID, "error", err)
 				time.Sleep(1 * time.Second)
 				continue
 			}
@@ -104,7 +128,7 @@ func (p *WorkerPool) workerLoop(workerID int) {
 				continue
 			}
 
-			logger.Log.Debugf("[Worker %d] Picked up task %d (%s)", workerID, task.ID, task.TaskType)
+			logger.Log.Debugw("Picked up task", "component", "Worker", "worker_id", workerID, "task_id", task.ID, "task_type", task.TaskType)
 			
 			// Process task based on type
 			var processErr error
@@ -117,15 +141,17 @@ func (p *WorkerPool) workerLoop(workerID int) {
 				processErr = p.handleProcess(p.ctx, task)
 			case "ANALYZE":
 				processErr = p.handleAnalyze(p.ctx, task)
+			case "EXPIRE_JOB":
+				processErr = p.handleExpireJob(p.ctx, task)
 			default:
 				processErr = fmt.Errorf("unknown task type: %s", task.TaskType)
 			}
 
 			if processErr != nil {
-				logger.Log.Errorf("[Worker %d] Task %d FAILED: %v", workerID, task.ID, processErr)
+				logger.Log.Errorw("Task FAILED", "component", "Worker", "worker_id", workerID, "task_id", task.ID, "error", processErr)
 				_ = p.db.FailTask(task.ID, processErr.Error())
 			} else {
-				logger.Log.Debugf("[Worker %d] Task %d COMPLETED", workerID, task.ID)
+				logger.Log.Debugw("Task COMPLETED", "component", "Worker", "worker_id", workerID, "task_id", task.ID)
 				_ = p.db.CompleteTask(task.ID)
 			}
 		}
@@ -157,6 +183,11 @@ type AnalyzePayload struct {
 	RawJobID int64 `json:"raw_job_id"`
 }
 
+type ExpireJobPayload struct {
+	RawJobID int64  `json:"raw_job_id"`
+	Reason   string `json:"reason"`
+}
+
 // ---------------------------------------------------------
 // Task Handlers
 // ---------------------------------------------------------
@@ -177,6 +208,8 @@ func (p *WorkerPool) handleDiscover(ctx context.Context, task *db.Task) error {
 		// Check deduplication (already fetched/inserted jobs)
 		exists, err := p.db.CanonicalIDExists(disc.CanonicalID)
 		if err == nil && exists {
+			// Update last seen timestamp to confirm job is still active on target board
+			_ = p.db.TouchJobLastSeen(disc.CanonicalID)
 			continue // Skip existing job
 		}
 
@@ -241,14 +274,21 @@ func (p *WorkerPool) handleProcess(ctx context.Context, task *db.Task) error {
 
 	keep, score, reason := p.filterPipe.Execute(ctx, &job.RawJob)
 	if !keep {
-		// Create a rejected JobAnalysis record. Store reason in missing_highlights JSON array
+		// Disqualified by blacklist: create a rejected record
 		reasonJSON, _ := json.Marshal([]string{reason})
-		_ = p.db.CreateJobAnalysisFull(payload.RawJobID, score, false, "[]", string(reasonJSON), "[]", "PARSED")
+		_ = p.db.CreateJobAnalysisFull(payload.RawJobID, 0.0, 0.0, false, "Disqualified: "+reason, "[]", string(reasonJSON), "[]", "REJECTED")
 		return nil
 	}
 
-	// It's qualified for analysis. Create the base record and enqueue it.
-	_ = p.db.CreateJobAnalysis(payload.RawJobID, score, true, "PENDING")
+	// Gating threshold check
+	if score < p.analysisThreshold {
+		summary := fmt.Sprintf("Keyword fit (%.0f) below threshold (%.0f)", score, p.analysisThreshold)
+		_ = p.db.CreateJobAnalysisFull(payload.RawJobID, score, score, false, summary, "[]", "[]", "[]", "FILTERED_OUT")
+		return nil
+	}
+
+	// Qualified for deep LLM analysis. Create the pending record and enqueue ANALYZE task.
+	_ = p.db.CreateJobAnalysis(payload.RawJobID, score, score, true, "PENDING")
 
 	analyzePayload := AnalyzePayload(payload)
 	_ = p.db.EnqueueTask("ANALYZE", analyzePayload)
@@ -267,9 +307,16 @@ func (p *WorkerPool) handleAnalyze(ctx context.Context, task *db.Task) error {
 		return err
 	}
 
-	// Use empty matched/missing keywords for now, or pass from filter if stored
 	result, err := p.analyzer.AnalyzeJD(ctx, job.Title, job.RawContent)
 	if err != nil {
+		logger.Log.Errorw("LLM analysis failed", "component", "Worker", "raw_job_id", payload.RawJobID, "error", err)
+		filterScore := 0.0
+		isQual := false
+		if job.Analysis != nil {
+			filterScore = job.Analysis.FilterScore
+			isQual = job.Analysis.IsQualified
+		}
+		_ = p.db.UpdateJobAnalysisFull(payload.RawJobID, filterScore, filterScore, isQual, "Analysis error: "+err.Error(), "[]", "[]", "[]", "FAILED")
 		return err
 	}
 
@@ -279,6 +326,9 @@ func (p *WorkerPool) handleAnalyze(ctx context.Context, task *db.Task) error {
 
 	err = p.db.UpdateJobAnalysisLLM(
 		payload.RawJobID,
+		result.AIScore,
+		result.IsQualified,
+		result.Summary,
 		string(matchedStr),
 		string(missingStr),
 		string(tailoringStr),
@@ -287,10 +337,55 @@ func (p *WorkerPool) handleAnalyze(ctx context.Context, task *db.Task) error {
 	return err
 }
 
+func (p *WorkerPool) handleExpireJob(ctx context.Context, task *db.Task) error {
+	var payload ExpireJobPayload
+	if err := json.Unmarshal([]byte(task.Payload), &payload); err != nil {
+		return err
+	}
+	logger.Log.Infow("Expiring job", "component", "Worker", "raw_job_id", payload.RawJobID, "reason", payload.Reason)
+	return p.db.ExpireJob(payload.RawJobID, payload.Reason)
+}
+
 func (p *WorkerPool) getCrawlerForPlatform(platform string) crawler.SiteCrawler {
 	plat := strings.ToLower(platform)
 	if strings.Contains(plat, "google") {
 		return google.NewGoogleCrawler()
 	}
+	if strings.Contains(plat, "apple") {
+		return apple.NewAppleCrawler()
+	}
+	if strings.Contains(plat, "meta") || strings.Contains(plat, "facebook") {
+		return meta.NewMetaCrawler()
+	}
+	if strings.Contains(plat, "amazon") || strings.Contains(plat, "aws") {
+		return amazon.NewAmazonCrawler()
+	}
+	if strings.Contains(plat, "microsoft") || strings.Contains(plat, "azure") {
+		return microsoft.NewMicrosoftCrawler()
+	}
+	if strings.Contains(plat, "bytedance") || strings.Contains(plat, "tiktok") {
+		return bytedance.NewByteDanceCrawler()
+	}
+	if strings.Contains(plat, "nvidia") {
+		return nvidia.NewNvidiaCrawler()
+	}
+	if strings.Contains(plat, "netflix") {
+		return netflix.NewNetflixCrawler()
+	}
+	if strings.Contains(plat, "uber") {
+		return uber.NewUberCrawler()
+	}
+	if strings.Contains(plat, "linkedin") {
+		return linkedin.NewLinkedInCrawler()
+	}
+	if strings.Contains(plat, "stripe") {
+		return stripe.NewStripeCrawler()
+	}
+	if strings.Contains(plat, "databricks") {
+		return databricks.NewDatabricksCrawler()
+	}
+	if strings.Contains(plat, "snowflake") {
+		return snowflake.NewSnowflakeCrawler()
+	}
 	return crawler.NewGenericCrawler()
 }

pkg/crawler/amazon/amazon.go

@@ -0,0 +1,285 @@
+package amazon
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type AmazonCrawler struct {
+	client *http.Client
+}
+
+func NewAmazonCrawler() *AmazonCrawler {
+	return NewAmazonCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewAmazonCrawlerWithClient(client *http.Client) *AmazonCrawler {
+	return &AmazonCrawler{
+		client: client,
+	}
+}
+
+func (c *AmazonCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "amazon" || p == "aws" || strings.Contains(p, "amazon.jobs")
+}
+
+type amazonSearchResponse struct {
+	Hits int             `json:"hits"`
+	Jobs []amazonJobItem `json:"jobs"`
+}
+
+type amazonJobItem struct {
+	ID                  string `json:"id"`
+	IDIsp               string `json:"id_icims"`
+	Title               string `json:"title"`
+	Company             string `json:"company_name"`
+	Location            string `json:"location"`
+	City                string `json:"city"`
+	State               string `json:"state"`
+	Country             string `json:"country_code"`
+	JobPath             string `json:"job_path"`
+	Description         string `json:"description"`
+	BasicQualifications string `json:"basic_qualifications"`
+	PreferredQuals      string `json:"preferred_qualifications"`
+	JobCategory         string `json:"job_category"`
+	JobFamily           string `json:"job_family_name"`
+	PostedDate          string `json:"posted_date"`
+}
+
+func (c *AmazonCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	searchAPI := targetURL
+	if !strings.Contains(searchAPI, "search.json") {
+		u, err := url.Parse(targetURL)
+		if err == nil && u.Host != "" {
+			q := u.Query()
+			searchAPI = fmt.Sprintf("https://www.amazon.jobs/en/search.json?%s", q.Encode())
+		} else {
+			searchAPI = fmt.Sprintf("https://www.amazon.jobs/en/search.json?base_query=%s&loc_query=Seattle%%2C+WA%%2C+USA", url.QueryEscape(targetURL))
+		}
+	}
+
+	resultLimit := 100
+	for page := 0; page < 5; page++ {
+		offset := page * resultLimit
+		reqURL := searchAPI
+		if strings.Contains(reqURL, "?") {
+			reqURL = fmt.Sprintf("%s&result_limit=%d&offset=%d", searchAPI, resultLimit, offset)
+		} else {
+			reqURL = fmt.Sprintf("%s?result_limit=%d&offset=%d", searchAPI, resultLimit, offset)
+		}
+
+		body, err := c.fetchURL(ctx, reqURL)
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+
+		var resp amazonSearchResponse
+		if err := json.Unmarshal([]byte(body), &resp); err != nil {
+			if page == 0 {
+				return nil, fmt.Errorf("failed to parse Amazon search response: %w", err)
+			}
+			break
+		}
+
+		if len(resp.Jobs) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, job := range resp.Jobs {
+			jobID := job.ID
+			if jobID == "" {
+				jobID = job.IDIsp
+			}
+			if jobID == "" {
+				continue
+			}
+
+			applyURL := fmt.Sprintf("https://www.amazon.jobs/en/jobs/%s", jobID)
+			if job.JobPath != "" {
+				if strings.HasPrefix(job.JobPath, "http") {
+					applyURL = job.JobPath
+				} else {
+					applyURL = "https://www.amazon.jobs" + job.JobPath
+				}
+			}
+
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("amazon:%s", jobID),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(resp.Jobs) < resultLimit {
+			break
+		}
+	}
+
+	logger.Log.Infow("Amazon discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *AmazonCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseAmazonJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	jsonURL := fmt.Sprintf("https://www.amazon.jobs/en/jobs/%s.json", jobID)
+	body, err := c.fetchURL(ctx, jsonURL)
+	if err == nil && json.Valid([]byte(body)) {
+		var detail struct {
+			Job amazonJobItem `json:"job"`
+		}
+		if err := json.Unmarshal([]byte(body), &detail); err == nil && detail.Job.Title != "" {
+			j := detail.Job
+			var sb strings.Builder
+			if j.Description != "" {
+				sb.WriteString("## Description\n\n")
+				sb.WriteString(cleanHTML(j.Description))
+				sb.WriteString("\n\n")
+			}
+			if j.BasicQualifications != "" {
+				sb.WriteString("## Basic Qualifications\n\n")
+				sb.WriteString(cleanHTML(j.BasicQualifications))
+				sb.WriteString("\n\n")
+			}
+			if j.PreferredQuals != "" {
+				sb.WriteString("## Preferred Qualifications\n\n")
+				sb.WriteString(cleanHTML(j.PreferredQuals))
+				sb.WriteString("\n\n")
+			}
+
+			company := "Amazon"
+			if j.Company != "" {
+				company = j.Company
+			}
+			loc := j.Location
+			if loc == "" && j.City != "" {
+				loc = fmt.Sprintf("%s, %s, %s", j.City, j.State, j.Country)
+			}
+			if loc == "" {
+				loc = "Seattle, WA, USA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("amazon:%s", jobID),
+				Title:       j.Title,
+				Company:     company,
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  sb.String(),
+			}, nil
+		}
+	}
+
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*class="[^"]*title[^"]*"[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	}
+	if title == "" {
+		title = "Software Engineer - Amazon"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("amazon:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Amazon",
+		Location:    "Seattle, WA, USA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *AmazonCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseAmazonJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/([0-9A-Za-z\-]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return strings.TrimSuffix(m[1], ".json")
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/amazon/amazon_test.go

@@ -0,0 +1,111 @@
+package amazon
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestAmazonCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "search.json") {
+				body := `{
+					"hits": 1,
+					"jobs": [
+						{
+							"id": "2500123",
+							"title": "Software Development Engineer II - AWS",
+							"company_name": "Amazon Web Services, Inc.",
+							"city": "Seattle",
+							"state": "WA",
+							"country_code": "USA",
+							"job_path": "/en/jobs/2500123/software-development-engineer-ii-aws"
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "2500123.json") {
+				body := `{
+					"job": {
+						"id": "2500123",
+						"title": "Software Development Engineer II - AWS",
+						"company_name": "Amazon Web Services, Inc.",
+						"location": "Seattle, WA, USA",
+						"description": "<p>Come build the future of cloud computing.</p>",
+						"basic_qualifications": "<li>3+ years of non-internship professional software development experience</li>",
+						"preferred_qualifications": "<li>Experience with distributed systems and Go/Java/C++</li>"
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewAmazonCrawlerWithClient(mockClient)
+	if !c.CanHandle("amazon") || !c.CanHandle("aws") {
+		t.Errorf("CanHandle should support amazon and aws")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://www.amazon.jobs/en/search.json?base_query=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "amazon:2500123" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://www.amazon.jobs/en/jobs/2500123")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Development Engineer II - AWS" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Amazon Web Services, Inc." {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+}
+
+func TestAmazonCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Amazon",
+		Crawler:                 NewAmazonCrawler(),
+		TargetURL:               "https://www.amazon.jobs/en/search.json?base_query=Software+Development+Engineer&loc_query=Seattle%2C+WA%2C+USA",
+		ExpectedCanonicalPrefix: "amazon:",
+		ExpectedCompany:         "Amazon",
+	})
+}

pkg/crawler/apple/apple.go

@@ -0,0 +1,250 @@
+package apple
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type AppleCrawler struct {
+	client *http.Client
+}
+
+func NewAppleCrawler() *AppleCrawler {
+	return &AppleCrawler{
+		client: &http.Client{
+			Timeout: 30 * time.Second,
+		},
+	}
+}
+
+func (a *AppleCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "apple" || p == "apple_jobs" || strings.Contains(p, "jobs.apple.com")
+}
+
+func (a *AppleCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	// Regex to match <a ... href="/en-us/details/200677768...">
+	reID := regexp.MustCompile(`href="(/[^"]*/details/(\d+)[^"]*)"`)
+
+	for page := 1; page <= 10; page++ {
+		pageURL := targetURL
+		if strings.Contains(pageURL, "?") {
+			pageURL = fmt.Sprintf("%s&page=%d", targetURL, page)
+		} else {
+			pageURL = fmt.Sprintf("%s?page=%d", targetURL, page)
+		}
+
+		htmlContent, err := a.fetchURL(ctx, pageURL)
+		if err != nil {
+			if page == 1 {
+				return nil, err
+			}
+			break
+		}
+
+		matches := reID.FindAllStringSubmatch(htmlContent, -1)
+		if len(matches) == 0 {
+			// No more jobs found on this page
+			break
+		}
+
+		newJobsFound := false
+		for _, m := range matches {
+			if len(m) >= 3 {
+				href := m[1]
+				jobID := m[2]
+				
+				applyURL := href
+				if strings.HasPrefix(applyURL, "/") {
+					applyURL = "https://jobs.apple.com" + applyURL
+				}
+				
+				if !seen[applyURL] {
+					seen[applyURL] = true
+					newJobsFound = true
+					discovered = append(discovered, crawler.DiscoveredJob{
+						CanonicalID: fmt.Sprintf("apple:%s", jobID),
+						ApplyURL:    applyURL,
+					})
+				}
+			}
+		}
+		
+		if !newJobsFound {
+			break
+		}
+	}
+
+	return discovered, nil
+}
+
+func (a *AppleCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	htmlContent, err := a.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	jobID := parseIDFromURL(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	// Extract window.__staticRouterHydrationData
+	re := regexp.MustCompile(`window\.__staticRouterHydrationData\s*=\s*JSON\.parse\("(.*?)"\);`)
+	match := re.FindStringSubmatch(htmlContent)
+	if len(match) < 2 {
+		return nil, fmt.Errorf("could not find JSON payload in Apple job page")
+	}
+
+	quotedStr := "\"" + match[1] + "\""
+	var rawJson string
+	if err := json.Unmarshal([]byte(quotedStr), &rawJson); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal JSON string literal: %w", err)
+	}
+
+	var payload map[string]interface{}
+	if err := json.Unmarshal([]byte(rawJson), &payload); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal JSON payload: %w", err)
+	}
+
+	var jobDetails map[string]interface{}
+	loaderData, ok := payload["loaderData"].(map[string]interface{})
+	if ok {
+		if jdWrapper, ok := loaderData["jobDetails"].(map[string]interface{}); ok {
+			if jd, ok := jdWrapper["jobsData"].(map[string]interface{}); ok {
+				jobDetails = jd
+			}
+		}
+	}
+
+	if jobDetails == nil {
+		return nil, fmt.Errorf("could not find jobsData in the parsed JSON")
+	}
+
+	title := getString(jobDetails, "postingTitle", "Unknown Apple Position")
+	company := "Apple"
+	
+	// Try to get location
+	var locationParts []string
+	if locs, ok := jobDetails["locations"].([]interface{}); ok {
+		for _, loc := range locs {
+			if locMap, ok := loc.(map[string]interface{}); ok {
+				if name, ok := locMap["name"].(string); ok && name != "" {
+					locationParts = append(locationParts, name)
+				}
+			}
+		}
+	}
+	location := "Apple Location"
+	if len(locationParts) > 0 {
+		location = strings.Join(locationParts, "; ")
+	}
+
+	// Construct raw markdown text
+	var sb strings.Builder
+	
+	if summary := getString(jobDetails, "jobSummary", ""); summary != "" {
+		sb.WriteString("## Summary\n\n")
+		sb.WriteString(cleanHTML(summary))
+		sb.WriteString("\n\n")
+	}
+	if desc := getString(jobDetails, "description", ""); desc != "" {
+		sb.WriteString("## Description\n\n")
+		sb.WriteString(cleanHTML(desc))
+		sb.WriteString("\n\n")
+	}
+	if minQual := getString(jobDetails, "minimumQualifications", ""); minQual != "" {
+		sb.WriteString("## Minimum Qualifications\n\n")
+		sb.WriteString(cleanHTML(minQual))
+		sb.WriteString("\n\n")
+	}
+	if prefQual := getString(jobDetails, "preferredQualifications", ""); prefQual != "" {
+		sb.WriteString("## Preferred Qualifications\n\n")
+		sb.WriteString(cleanHTML(prefQual))
+		sb.WriteString("\n\n")
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("apple:%s", jobID),
+		Title:       title,
+		Company:     company,
+		Location:    location,
+		ApplyURL:    applyURL,
+		RawContent:  sb.String(),
+	}, nil
+}
+
+func (a *AppleCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create HTTP request for %s: %w", urlStr, err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8")
+
+	resp, err := a.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP fetch failed for %s: %w", urlStr, err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		logger.Log.Warnw("Apple crawler HTTP error", "status", resp.StatusCode, "url", urlStr)
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response body for %s: %w", urlStr, err)
+	}
+
+	return string(body), nil
+}
+
+func parseIDFromURL(u string) string {
+	re := regexp.MustCompile(`/details/(\d+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func getString(m map[string]interface{}, key, fallback string) string {
+	if val, ok := m[key]; ok {
+		if s, ok := val.(string); ok && s != "" {
+			return s
+		}
+	}
+	return fallback
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	return strings.TrimSpace(s)
+}

pkg/crawler/apple/apple_test.go

@@ -0,0 +1,67 @@
+package apple
+
+import (
+	"context"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestAppleCrawler_EndToEnd(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping end-to-end network test in short mode")
+	}
+
+	crawler := NewAppleCrawler()
+
+	// Use the default test URL
+	targetURL := "https://jobs.apple.com/en-us/search?location=seattle-SEA&key=Software%2520Engineer"
+
+	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+	defer cancel()
+
+	// 1. Test DiscoverJobs
+	t.Logf("Discovering jobs from: %s", targetURL)
+	jobs, err := crawler.DiscoverJobs(ctx, targetURL)
+	if err != nil {
+		t.Logf("Notice: Network fetch error (expected in offline sandbox): %v", err)
+		return
+	}
+
+	if len(jobs) == 0 {
+		t.Fatalf("DiscoverJobs returned 0 jobs for URL: %s", targetURL)
+	}
+	t.Logf("Successfully discovered %d jobs", len(jobs))
+
+	// 2. Pick the first job and test FetchJobDetail
+	testJob := jobs[0]
+	t.Logf("Fetching details for job: %s", testJob.ApplyURL)
+
+	details, err := crawler.FetchJobDetail(ctx, testJob.ApplyURL)
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if details == nil {
+		t.Fatal("FetchJobDetail returned nil ExtractedJob")
+	}
+
+	// 3. Verify extracted details
+	t.Logf("Extracted Title: %s", details.Title)
+	t.Logf("Extracted Company: %s", details.Company)
+	t.Logf("Extracted Location: %s", details.Location)
+	t.Logf("Raw Content Length: %d", len(details.RawContent))
+
+	if details.Title == "" || details.Title == "Unknown Apple Position" {
+		t.Errorf("Expected valid job title, got %q", details.Title)
+	}
+	if details.Company != "Apple" {
+		t.Errorf("Expected company 'Apple', got %q", details.Company)
+	}
+	if !strings.HasPrefix(details.CanonicalID, "apple:") {
+		t.Errorf("Expected CanonicalID to start with 'apple:', got %q", details.CanonicalID)
+	}
+	if len(details.RawContent) < 100 {
+		t.Errorf("Expected RawContent to be substantive, got length %d", len(details.RawContent))
+	}
+}

pkg/crawler/bytedance/bytedance.go

@@ -0,0 +1,272 @@
+package bytedance
+
+import (
+	"bytes"
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type ByteDanceCrawler struct {
+	client *http.Client
+}
+
+func NewByteDanceCrawler() *ByteDanceCrawler {
+	return NewByteDanceCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewByteDanceCrawlerWithClient(client *http.Client) *ByteDanceCrawler {
+	return &ByteDanceCrawler{
+		client: client,
+	}
+}
+
+func (c *ByteDanceCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "bytedance" || p == "tiktok" || strings.Contains(p, "bytedance.com") || strings.Contains(p, "tiktok.com")
+}
+
+type byteDanceSearchPayload struct {
+	Keyword       string   `json:"keyword"`
+	Limit         int      `json:"limit"`
+	Offset        int      `json:"offset"`
+	JobCategoryID []string `json:"job_category_id_list,omitempty"`
+	LocationList  []string `json:"location_code_list,omitempty"`
+}
+
+type byteDanceSearchResponse struct {
+	Code int `json:"code"`
+	Data struct {
+		Count    int                `json:"count"`
+		JobPosts []byteDanceJobItem `json:"job_post_list"`
+	} `json:"data"`
+}
+
+type byteDanceDetailResponse struct {
+	Code int `json:"code"`
+	Data struct {
+		JobPost byteDanceJobItem `json:"job_post"`
+	} `json:"data"`
+}
+
+type byteDanceJobItem struct {
+	ID          string `json:"id"`
+	Title       string `json:"title"`
+	Description string `json:"description"`
+	Requirement string `json:"requirement"`
+	CityInfo    struct {
+		Name     string `json:"name"`
+		EnName   string `json:"en_name"`
+		Location string `json:"location_code"`
+	} `json:"city_info"`
+	JobCategory struct {
+		Name string `json:"name"`
+	} `json:"job_category"`
+	PublishTime int64 `json:"publish_time"`
+}
+
+func (c *ByteDanceCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	// Default API endpoint
+	apiURL := "https://jobs.bytedance.com/api/v1/search/job/posts"
+	if strings.Contains(targetURL, "tiktok.com") {
+		apiURL = "https://careers.tiktok.com/api/v1/search/job/posts"
+	}
+
+	keyword := "Software Engineer"
+	u, err := url.Parse(targetURL)
+	if err == nil && u.Query().Get("keyword") != "" {
+		keyword = u.Query().Get("keyword")
+	}
+
+	pageSize := 50
+	for page := 0; page < 5; page++ {
+		offset := page * pageSize
+		reqPayload := byteDanceSearchPayload{
+			Keyword: keyword,
+			Limit:   pageSize,
+			Offset:  offset,
+		}
+
+		payloadBytes, err := json.Marshal(reqPayload)
+		if err != nil {
+			return nil, err
+		}
+
+		req, err := http.NewRequestWithContext(ctx, "POST", apiURL, bytes.NewBuffer(payloadBytes))
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+
+		req.Header.Set("Content-Type", "application/json")
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+
+		resp, err := c.client.Do(req)
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+		defer func() { _ = resp.Body.Close() }()
+
+		body, err := io.ReadAll(resp.Body)
+		if err != nil {
+			break
+		}
+
+		var searchResp byteDanceSearchResponse
+		if err := json.Unmarshal(body, &searchResp); err != nil {
+			if page == 0 {
+				return nil, fmt.Errorf("failed to parse ByteDance search response: %w", err)
+			}
+			break
+		}
+
+		posts := searchResp.Data.JobPosts
+		if len(posts) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, post := range posts {
+			if post.ID == "" {
+				continue
+			}
+
+			applyURL := fmt.Sprintf("https://jobs.bytedance.com/en/position/%s/detail", post.ID)
+			if strings.Contains(targetURL, "tiktok.com") {
+				applyURL = fmt.Sprintf("https://careers.tiktok.com/position/%s/detail", post.ID)
+			}
+
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("bytedance:%s", post.ID),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(posts) < pageSize {
+			break
+		}
+	}
+
+	logger.Log.Infow("ByteDance discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *ByteDanceCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseByteDanceJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	apiURL := fmt.Sprintf("https://jobs.bytedance.com/api/v1/job/posts/%s", jobID)
+	if strings.Contains(applyURL, "tiktok.com") {
+		apiURL = fmt.Sprintf("https://careers.tiktok.com/api/v1/job/posts/%s", jobID)
+	}
+
+	req, err := http.NewRequestWithContext(ctx, "GET", apiURL, nil)
+	if err == nil {
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+		resp, err := c.client.Do(req)
+		if err == nil {
+			defer func() { _ = resp.Body.Close() }()
+			body, _ := io.ReadAll(resp.Body)
+			var detailResp byteDanceDetailResponse
+			if err := json.Unmarshal(body, &detailResp); err == nil && detailResp.Data.JobPost.Title != "" {
+				p := detailResp.Data.JobPost
+				var sb strings.Builder
+				if p.Description != "" {
+					sb.WriteString("## Responsibilities\n\n")
+					sb.WriteString(cleanHTML(p.Description))
+					sb.WriteString("\n\n")
+				}
+				if p.Requirement != "" {
+					sb.WriteString("## Qualifications\n\n")
+					sb.WriteString(cleanHTML(p.Requirement))
+					sb.WriteString("\n\n")
+				}
+
+				loc := p.CityInfo.EnName
+				if loc == "" {
+					loc = p.CityInfo.Name
+				}
+				if loc == "" {
+					loc = "Seattle / Bellevue, WA"
+				}
+
+				company := "ByteDance"
+				if strings.Contains(applyURL, "tiktok") {
+					company = "TikTok"
+				}
+
+				return &crawler.ExtractedJob{
+					CanonicalID: fmt.Sprintf("bytedance:%s", jobID),
+					Title:       p.Title,
+					Company:     company,
+					Location:    loc,
+					ApplyURL:    applyURL,
+					RawContent:  sb.String(),
+				}, nil
+			}
+		}
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("bytedance:%s", jobID),
+		Title:       "Software Engineer - ByteDance",
+		Company:     "ByteDance",
+		Location:    "Seattle / Bellevue, WA",
+		ApplyURL:    applyURL,
+		RawContent:  "See job posting at " + applyURL,
+	}, nil
+}
+
+func parseByteDanceJobID(u string) string {
+	re := regexp.MustCompile(`/position/([0-9A-Za-z]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/bytedance/bytedance_test.go

@@ -0,0 +1,119 @@
+package bytedance
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestByteDanceCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/search/job/posts") {
+				body := `{
+					"code": 0,
+					"data": {
+						"count": 1,
+						"job_post_list": [
+							{
+								"id": "7198234123",
+								"title": "Software Engineer - Distributed Storage (Seattle/Bellevue)",
+								"city_info": {
+									"name": "Bellevue",
+									"en_name": "Bellevue, WA"
+								}
+							}
+						]
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "/job/posts/7198234123") {
+				body := `{
+					"code": 0,
+					"data": {
+						"job_post": {
+							"id": "7198234123",
+							"title": "Software Engineer - Distributed Storage (Seattle/Bellevue)",
+							"description": "<p>Build high-throughput storage engines for TikTok recommendation algorithms.</p>",
+							"requirement": "<li>BS/MS/PhD in CS and 3+ years experience with C++/Go.</li>",
+							"city_info": {
+								"en_name": "Bellevue, WA"
+							}
+						}
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewByteDanceCrawlerWithClient(mockClient)
+	if !c.CanHandle("bytedance") || !c.CanHandle("tiktok") {
+		t.Errorf("CanHandle should support bytedance and tiktok")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://jobs.bytedance.com/api/v1/search/job/posts?keyword=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "bytedance:7198234123" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://jobs.bytedance.com/en/position/7198234123/detail")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer - Distributed Storage (Seattle/Bellevue)" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "ByteDance" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Bellevue, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestByteDanceCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "ByteDance",
+		Crawler:                 NewByteDanceCrawler(),
+		TargetURL:               "https://jobs.bytedance.com/api/v1/search/job/posts?keyword=Software+Engineer",
+		ExpectedCanonicalPrefix: "bytedance:",
+		ExpectedCompany:         "ByteDance",
+	})
+}

pkg/crawler/databricks/databricks.go

@@ -0,0 +1,225 @@
+package databricks
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type DatabricksCrawler struct {
+	client *http.Client
+}
+
+func NewDatabricksCrawler() *DatabricksCrawler {
+	return NewDatabricksCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewDatabricksCrawlerWithClient(client *http.Client) *DatabricksCrawler {
+	return &DatabricksCrawler{
+		client: client,
+	}
+}
+
+func (c *DatabricksCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "databricks" || strings.Contains(p, "databricks.com/company/careers") || strings.Contains(p, "boards.greenhouse.io/databricks")
+}
+
+type databricksJobsResponse struct {
+	Jobs []databricksJob `json:"jobs"`
+}
+
+type databricksJob struct {
+	ID          int64  `json:"id"`
+	Title       string `json:"title"`
+	AbsoluteURL string `json:"absolute_url"`
+	Location    struct {
+		Name string `json:"name"`
+	} `json:"location"`
+	Offices []struct {
+		Name     string `json:"name"`
+		Location string `json:"location"`
+	} `json:"offices"`
+	Departments []struct {
+		Name string `json:"name"`
+	} `json:"departments"`
+	Content string `json:"content"`
+}
+
+func (c *DatabricksCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	apiURL := "https://boards-api.greenhouse.io/v1/boards/databricks/jobs"
+	if targetURL != "" && strings.Contains(targetURL, "boards-api.greenhouse.io") {
+		apiURL = targetURL
+	}
+
+	body, err := c.fetchURL(ctx, apiURL)
+	if err != nil {
+		return nil, err
+	}
+
+	var resp databricksJobsResponse
+	if err := json.Unmarshal([]byte(body), &resp); err != nil {
+		return nil, fmt.Errorf("failed to parse Databricks Greenhouse API response: %w", err)
+	}
+
+	for _, job := range resp.Jobs {
+		if job.ID == 0 {
+			continue
+		}
+
+		jobIDStr := fmt.Sprintf("%d", job.ID)
+		applyURL := job.AbsoluteURL
+		if applyURL == "" {
+			applyURL = fmt.Sprintf("https://boards.greenhouse.io/databricks/jobs/%s", jobIDStr)
+		}
+
+		if !seen[applyURL] {
+			seen[applyURL] = true
+			discovered = append(discovered, crawler.DiscoveredJob{
+				CanonicalID: fmt.Sprintf("databricks:%s", jobIDStr),
+				ApplyURL:    applyURL,
+			})
+		}
+	}
+
+	logger.Log.Infow("Databricks discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *DatabricksCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseDatabricksJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	detailAPI := fmt.Sprintf("https://boards-api.greenhouse.io/v1/boards/databricks/jobs/%s", jobID)
+	body, err := c.fetchURL(ctx, detailAPI)
+	if err == nil && json.Valid([]byte(body)) {
+		var job databricksJob
+		if err := json.Unmarshal([]byte(body), &job); err == nil && job.Title != "" {
+			loc := job.Location.Name
+			if loc == "" && len(job.Offices) > 0 {
+				var locs []string
+				for _, o := range job.Offices {
+					if o.Name != "" {
+						locs = append(locs, o.Name)
+					}
+				}
+				loc = strings.Join(locs, "; ")
+			}
+			if loc == "" {
+				loc = "Bellevue / Seattle, WA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("databricks:%s", jobID),
+				Title:       job.Title,
+				Company:     "Databricks",
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  cleanHTML(job.Content),
+			}, nil
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*class="app-title"[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	}
+	if title == "" {
+		title = "Software Engineer - Databricks"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("databricks:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Databricks",
+		Location:    "Bellevue / Seattle, WA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *DatabricksCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseDatabricksJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/(\d+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/databricks/databricks_test.go

@@ -0,0 +1,109 @@
+package databricks
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestDatabricksCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.HasSuffix(r.URL.Path, "/jobs/5192301") {
+				body := `{
+					"id": 5192301,
+					"title": "Staff Software Engineer - Lakehouse Query Engine",
+					"content": "<p>Databricks is creating the world's most performant unified analytics engine.</p>",
+					"location": {
+						"name": "Bellevue, WA"
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/jobs") {
+				body := `{
+					"jobs": [
+						{
+							"id": 5192301,
+							"title": "Staff Software Engineer - Lakehouse Query Engine",
+							"absolute_url": "https://boards.greenhouse.io/databricks/jobs/5192301",
+							"location": {
+								"name": "Bellevue, WA"
+							}
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewDatabricksCrawlerWithClient(mockClient)
+	if !c.CanHandle("databricks") {
+		t.Errorf("CanHandle should support databricks")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://boards-api.greenhouse.io/v1/boards/databricks/jobs")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "databricks:5192301" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://boards.greenhouse.io/databricks/jobs/5192301")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Staff Software Engineer - Lakehouse Query Engine" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Databricks" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Bellevue, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestDatabricksCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Databricks",
+		Crawler:                 NewDatabricksCrawler(),
+		TargetURL:               "https://boards-api.greenhouse.io/v1/boards/databricks/jobs",
+		ExpectedCanonicalPrefix: "databricks:",
+		ExpectedCompany:         "Databricks",
+	})
+}

pkg/crawler/google/google.go

@@ -7,14 +7,14 @@ import (
 	"encoding/json"
 	"fmt"
 	"io"
-	"log"
 	"net/http"
 	"regexp"
 	"strings"
 	"time"
 
 	md "github.com/JohannesKaufmann/html-to-markdown/v2"
-	"github.com/hireme/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
 	"golang.org/x/net/html"
 )
 
@@ -183,7 +183,7 @@ func extractFromAFData(htmlContent string) (title, company, location string) {
 	marker := `AF_initDataCallback({key: 'ds:0'`
 	idx := strings.Index(htmlContent, marker)
 	if idx == -1 {
-		log.Printf("[GoogleCrawler] AF_initDataCallback ds:0 not found")
+		logger.Log.Debugw("AF_initDataCallback ds:0 not found", "component", "GoogleCrawler")
 		return "", "", ""
 	}
 
@@ -191,7 +191,7 @@ func extractFromAFData(htmlContent string) (title, company, location string) {
 	dataMarker := "data:"
 	dataIdx := strings.Index(htmlContent[idx:], dataMarker)
 	if dataIdx == -1 {
-		log.Printf("[GoogleCrawler] 'data:' not found after ds:0 marker")
+		logger.Log.Debugw("'data:' not found after ds:0 marker", "component", "GoogleCrawler")
 		return "", "", ""
 	}
 	dataStart := idx + dataIdx + len(dataMarker)
@@ -199,14 +199,14 @@ func extractFromAFData(htmlContent string) (title, company, location string) {
 	// Use bracket counting to find the matching ] for the outer [
 	jsonStr := extractBalancedBracket(htmlContent, dataStart)
 	if jsonStr == "" {
-		log.Printf("[GoogleCrawler] Failed to extract balanced bracket for ds:0 data")
+		logger.Log.Debugw("Failed to extract balanced bracket for ds:0 data", "component", "GoogleCrawler")
 		return "", "", ""
 	}
 
 	// Parse the outer array: data = [[...job record...]]
 	var outerArray []json.RawMessage
 	if err := json.Unmarshal([]byte(jsonStr), &outerArray); err != nil {
-		log.Printf("[GoogleCrawler] JSON parse error for ds:0 outer: %v", err)
+		logger.Log.Debugw("JSON parse error for ds:0 outer", "component", "GoogleCrawler", "error", err)
 		return "", "", ""
 	}
 	if len(outerArray) == 0 {
@@ -216,7 +216,7 @@ func extractFromAFData(htmlContent string) (title, company, location string) {
 	// Parse inner job record array
 	var record []json.RawMessage
 	if err := json.Unmarshal(outerArray[0], &record); err != nil {
-		log.Printf("[GoogleCrawler] JSON parse error for job record: %v", err)
+		logger.Log.Debugw("JSON parse error for job record", "component", "GoogleCrawler", "error", err)
 		return "", "", ""
 	}
 
@@ -247,7 +247,7 @@ func extractFromAFData(htmlContent string) (title, company, location string) {
 		}
 	}
 
-	log.Printf("[GoogleCrawler] AF_initDataCallback parsed: title=%q company=%q location=%q", title, company, location)
+	logger.Log.Debugw("AF_initDataCallback parsed job", "component", "GoogleCrawler", "title", title, "company", company, "location", location)
 	return title, company, location
 }
 
@@ -309,7 +309,7 @@ func (g *GoogleCrawler) fetchWithRetry(ctx context.Context, urlStr string) (stri
 	for attempt := 0; attempt < 3; attempt++ {
 		if attempt > 0 {
 			delay := time.Duration(attempt) * 2 * time.Second
-			log.Printf("[GoogleCrawler] Retry #%d for %s after %v", attempt, urlStr, delay)
+			logger.Log.Warnw("Retrying request", "component", "GoogleCrawler", "attempt", attempt, "url", urlStr, "delay", delay)
 			time.Sleep(delay)
 		}
 

pkg/crawler/linkedin/linkedin.go

@@ -0,0 +1,229 @@
+package linkedin
+
+import (
+	"context"
+	"crypto/sha256"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type LinkedInCrawler struct {
+	client *http.Client
+}
+
+func NewLinkedInCrawler() *LinkedInCrawler {
+	return NewLinkedInCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewLinkedInCrawlerWithClient(client *http.Client) *LinkedInCrawler {
+	return &LinkedInCrawler{
+		client: client,
+	}
+}
+
+func (c *LinkedInCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "linkedin" || strings.Contains(p, "linkedin.com/jobs")
+}
+
+func (c *LinkedInCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	searchAPI := targetURL
+	if !strings.Contains(searchAPI, "jobs-guest/jobs/api/seeMoreJobPostings/search") {
+		u, err := url.Parse(targetURL)
+		if err == nil && u.Host != "" {
+			q := u.Query()
+			searchAPI = fmt.Sprintf("https://www.linkedin.com/jobs-guest/jobs/api/seeMoreJobPostings/search?%s", q.Encode())
+		} else {
+			searchAPI = fmt.Sprintf("https://www.linkedin.com/jobs-guest/jobs/api/seeMoreJobPostings/search?keywords=%s&location=Seattle%%2C+Washington%%2C+United+States", url.QueryEscape(targetURL))
+		}
+	}
+
+	pageSize := 25
+	reID := regexp.MustCompile(`(?:urn:li:jobPosting:(\d+)|/jobs/view/(?:[a-z0-9\-]+-)?(\d+))`)
+
+	for page := 0; page < 4; page++ {
+		start := page * pageSize
+		reqURL := searchAPI
+		if strings.Contains(reqURL, "?") {
+			reqURL = fmt.Sprintf("%s&start=%d", searchAPI, start)
+		} else {
+			reqURL = fmt.Sprintf("%s?start=%d", searchAPI, start)
+		}
+
+		body, err := c.fetchURL(ctx, reqURL)
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+
+		matches := reID.FindAllStringSubmatch(body, -1)
+		if len(matches) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, m := range matches {
+			jobID := m[1]
+			if jobID == "" && len(m) >= 3 {
+				jobID = m[2]
+			}
+			if jobID == "" {
+				continue
+			}
+
+			applyURL := fmt.Sprintf("https://www.linkedin.com/jobs/view/%s", jobID)
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("linkedin:%s", jobID),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(matches) < pageSize {
+			break
+		}
+	}
+
+	logger.Log.Infow("LinkedIn discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *LinkedInCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseLinkedInJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	// LinkedIn guest API endpoint for job details:
+	guestDetailURL := fmt.Sprintf("https://www.linkedin.com/jobs-guest/jobs/api/jobPosting/%s", jobID)
+	body, err := c.fetchURL(ctx, guestDetailURL)
+	if err != nil {
+		body, err = c.fetchURL(ctx, applyURL)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	title := extractRegex(body, `<h2[^>]*class="[^"]*top-card-layout__title[^"]*"[^>]*>(.*?)</h2>`)
+	if title == "" {
+		title = extractRegex(body, `<h1[^>]*>(.*?)</h1>`)
+	}
+	if title == "" {
+		title = "Software Engineer"
+	}
+
+	company := extractRegex(body, `<a[^>]*class="[^"]*topcard__org-name-link[^"]*"[^>]*>(.*?)</a>`)
+	if company == "" {
+		company = extractRegex(body, `<span[^>]*class="[^"]*topcard__flavor[^"]*"[^>]*>(.*?)</span>`)
+	}
+	if company == "" {
+		company = "LinkedIn"
+	}
+
+	location := extractRegex(body, `<span[^>]*class="[^"]*topcard__flavor--bullet[^"]*"[^>]*>(.*?)</span>`)
+	if location == "" {
+		location = "Seattle, WA, USA"
+	}
+
+	desc := extractRegex(body, `<div[^>]*class="[^"]*show-more-less-html__markup[^"]*"[^>]*>([\s\S]*?)</div>`)
+	if desc == "" {
+		desc = body
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("linkedin:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     cleanHTML(company),
+		Location:    cleanHTML(location),
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(desc),
+	}, nil
+}
+
+func (c *LinkedInCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8")
+	req.Header.Set("Accept-Language", "en-US,en;q=0.9")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseLinkedInJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/view/(?:[a-z0-9\-]+-)?(\d+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	re2 := regexp.MustCompile(`/jobPosting/(\d+)`)
+	m2 := re2.FindStringSubmatch(u)
+	if len(m2) >= 2 {
+		return m2[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/linkedin/linkedin_test.go

@@ -0,0 +1,104 @@
+package linkedin
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestLinkedInCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "seeMoreJobPostings") {
+				body := `<html>
+					<li data-row="1">
+						<div class="base-card" data-entity-urn="urn:li:jobPosting:3920192831">
+							<a class="base-card__full-link" href="https://www.linkedin.com/jobs/view/software-engineer-infra-at-linkedin-3920192831"></a>
+						</div>
+					</li>
+				</html>`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "3920192831") {
+				body := `<html>
+					<h2 class="top-card-layout__title">Staff Software Engineer - Distributed Graph</h2>
+					<a class="topcard__org-name-link">LinkedIn</a>
+					<span class="topcard__flavor--bullet">Bellevue, WA</span>
+					<div class="show-more-less-html__markup">
+						<p>LinkedIn graph systems scale to billions of edge connections.</p>
+					</div>
+				</html>`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewLinkedInCrawlerWithClient(mockClient)
+	if !c.CanHandle("linkedin") {
+		t.Errorf("CanHandle should support linkedin")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://www.linkedin.com/jobs-guest/jobs/api/seeMoreJobPostings/search?keywords=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "linkedin:3920192831" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://www.linkedin.com/jobs/view/3920192831")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Staff Software Engineer - Distributed Graph" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "LinkedIn" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Bellevue, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestLinkedInCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "LinkedIn",
+		Crawler:                 NewLinkedInCrawler(),
+		TargetURL:               "https://www.linkedin.com/jobs-guest/jobs/api/seeMoreJobPostings/search?keywords=Software+Engineer&location=Seattle%2C+Washington%2C+United+States",
+		ExpectedCanonicalPrefix: "linkedin:",
+		ExpectedCompany:         "",
+	})
+}

pkg/crawler/meta/meta.go

@@ -0,0 +1,142 @@
+package meta
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/chromedp/cdproto/cdp"
+	"github.com/chromedp/chromedp"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type MetaCrawler struct{}
+
+func NewMetaCrawler() *MetaCrawler {
+	return &MetaCrawler{}
+}
+
+func (c *MetaCrawler) CanHandle(platform string) bool {
+	return strings.ToLower(platform) == "meta" || strings.ToLower(platform) == "facebook"
+}
+
+func (c *MetaCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	opts := append(chromedp.DefaultExecAllocatorOptions[:],
+		chromedp.Flag("headless", true),
+		chromedp.Flag("no-sandbox", true),
+		chromedp.Flag("disable-gpu", true),
+		chromedp.Flag("disable-dev-shm-usage", true),
+	)
+
+	allocCtx, cancelAlloc := chromedp.NewExecAllocator(ctx, opts...)
+	defer cancelAlloc()
+
+	chromeCtx, cancelCtx := chromedp.NewContext(allocCtx)
+	defer cancelCtx()
+
+	chromeCtx, cancelTimeout := context.WithTimeout(chromeCtx, 30*time.Second)
+	defer cancelTimeout()
+
+	var nodes []*cdp.Node
+	logger.Log.Infow("Navigating to Meta careers search", "url", targetURL)
+
+	err := chromedp.Run(chromeCtx,
+		chromedp.Navigate(targetURL),
+		chromedp.Sleep(5*time.Second), // Wait for React and GraphQL to load
+		chromedp.Nodes(`a`, &nodes, chromedp.ByQueryAll),
+	)
+
+	if err != nil {
+		return nil, fmt.Errorf("chromedp navigation failed: %w", err)
+	}
+
+	var results []crawler.DiscoveredJob
+	seen := make(map[string]bool)
+
+	for _, n := range nodes {
+		href := n.AttributeValue("href")
+		// Meta uses /v2/jobs/..., /jobs/..., or /profile/job_details/...
+		if (strings.Contains(href, "/v2/jobs/") || strings.Contains(href, "/jobs/") || strings.Contains(href, "/profile/job_details/")) && !strings.Contains(href, "jobsearch") {
+			fullURL := href
+			if strings.HasPrefix(href, "/") {
+				fullURL = "https://www.metacareers.com" + href
+			}
+			
+			// Extract Canonical ID from URL
+			parts := strings.Split(strings.Trim(href, "/"), "/")
+			canonicalID := "unknown"
+			for i, p := range parts {
+				if (p == "jobs" || p == "job_details") && i+1 < len(parts) {
+					canonicalID = parts[i+1]
+					break
+				}
+			}
+
+			if !seen[fullURL] {
+				seen[fullURL] = true
+				results = append(results, crawler.DiscoveredJob{
+					ApplyURL:    fullURL,
+					CanonicalID: "meta_" + canonicalID,
+				})
+			}
+		}
+	}
+
+	logger.Log.Infow("Meta discovery finished", "found_count", len(results))
+	return results, nil
+}
+
+func (c *MetaCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	opts := append(chromedp.DefaultExecAllocatorOptions[:],
+		chromedp.Flag("headless", true),
+		chromedp.Flag("no-sandbox", true),
+		chromedp.Flag("disable-gpu", true),
+		chromedp.Flag("disable-dev-shm-usage", true),
+	)
+
+	allocCtx, cancelAlloc := chromedp.NewExecAllocator(ctx, opts...)
+	defer cancelAlloc()
+
+	chromeCtx, cancelCtx := chromedp.NewContext(allocCtx)
+	defer cancelCtx()
+
+	chromeCtx, cancelTimeout := context.WithTimeout(chromeCtx, 30*time.Second)
+	defer cancelTimeout()
+
+	logger.Log.Infow("Fetching Meta job detail", "url", applyURL)
+
+	var titleText string
+	var bodyText string
+
+	err := chromedp.Run(chromeCtx,
+		chromedp.Navigate(applyURL),
+		chromedp.Sleep(5*time.Second),
+		chromedp.Evaluate(`document.querySelector("h1") ? document.querySelector("h1").innerText : "Unknown Meta Title"`, &titleText),
+		chromedp.Evaluate(`document.body.innerText`, &bodyText),
+	)
+
+	if err != nil {
+		return nil, fmt.Errorf("chromedp navigation failed: %w", err)
+	}
+
+	// Extract ID
+	parts := strings.Split(strings.Trim(applyURL, "/"), "/")
+	canonicalID := "unknown"
+	for i, p := range parts {
+		if (p == "jobs" || p == "job_details") && i+1 < len(parts) {
+			canonicalID = parts[i+1]
+			break
+		}
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: "meta_" + canonicalID,
+		Title:       titleText,
+		Company:     "Meta",
+		Location:    "Check description", // Location extraction is complex via text, leave it or parse later
+		ApplyURL:    applyURL,
+		RawContent:  bodyText,
+	}, nil
+}

pkg/crawler/meta/meta_test.go

@@ -0,0 +1,57 @@
+package meta
+
+import (
+	"context"
+	"testing"
+	"time"
+)
+
+func TestMetaCrawler_EndToEnd(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping long-running Meta crawler network test in short mode")
+	}
+
+	c := NewMetaCrawler()
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
+	defer cancel()
+
+	testURL := "https://www.metacareers.com/jobsearch/?offices[0]=Seattle%2C%20WA&roles[0]=Full%20time%20employment"
+
+	t.Logf("Discovering jobs from: %s", testURL)
+	discovered, err := c.DiscoverJobs(ctx, testURL)
+	if err != nil {
+		t.Logf("Notice: Chromedp execution error (expected in offline/sandboxed environment): %v", err)
+		return
+	}
+
+	if len(discovered) == 0 {
+		t.Fatalf("Expected to find at least 1 job, found 0")
+	}
+
+	t.Logf("Successfully discovered %d jobs", len(discovered))
+
+	// Fetch details for the first discovered job
+	firstJob := discovered[0]
+	t.Logf("Fetching details for job: %s", firstJob.ApplyURL)
+
+	detail, err := c.FetchJobDetail(ctx, firstJob.ApplyURL)
+	if err != nil {
+		t.Fatalf("Failed to fetch job details: %v", err)
+	}
+
+	if detail.Title == "" || detail.Title == "Unknown Meta Title" {
+		t.Errorf("Expected a valid job title, got %s", detail.Title)
+	}
+
+	if len(detail.RawContent) < 100 {
+		t.Errorf("Expected RawContent to contain the job description, but it was too short (%d bytes)", len(detail.RawContent))
+	}
+
+	if detail.CanonicalID == "unknown" || detail.CanonicalID == "" {
+		t.Errorf("Failed to extract canonical ID from %s", firstJob.ApplyURL)
+	}
+
+	t.Logf("Extracted Title: %s", detail.Title)
+	t.Logf("Extracted CanonicalID: %s", detail.CanonicalID)
+	t.Logf("Raw Content Length: %d", len(detail.RawContent))
+}

pkg/crawler/microsoft/microsoft.go

@@ -0,0 +1,275 @@
+package microsoft
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type MicrosoftCrawler struct {
+	client *http.Client
+}
+
+func NewMicrosoftCrawler() *MicrosoftCrawler {
+	return NewMicrosoftCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewMicrosoftCrawlerWithClient(client *http.Client) *MicrosoftCrawler {
+	return &MicrosoftCrawler{
+		client: client,
+	}
+}
+
+func (c *MicrosoftCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "microsoft" || p == "azure" || strings.Contains(p, "careers.microsoft.com")
+}
+
+type msSearchResponse struct {
+	OperationResult struct {
+		Result struct {
+			TotalJobs int        `json:"totalJobs"`
+			Jobs      []msJobItem `json:"jobs"`
+		} `json:"result"`
+	} `json:"operationResult"`
+}
+
+type msJobDetailResponse struct {
+	OperationResult struct {
+		Result msJobItem `json:"result"`
+	} `json:"operationResult"`
+}
+
+type msJobItem struct {
+	JobID          string   `json:"jobId"`
+	Title          string   `json:"title"`
+	Description    string   `json:"description"`
+	Qualifications string   `json:"qualifications"`
+	Responsibilities string `json:"responsibilities"`
+	PostingDate    string   `json:"postingDate"`
+	Properties     struct {
+		Locations          []string `json:"locations"`
+		PrimaryLocation    string   `json:"primaryLocation"`
+		WorkSiteFlexibility string  `json:"workSiteFlexibility"`
+		Profession         string   `json:"profession"`
+		RoleType           string   `json:"roleType"`
+		Discipline         string   `json:"discipline"`
+	} `json:"properties"`
+}
+
+func (c *MicrosoftCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	searchAPI := targetURL
+	if !strings.Contains(searchAPI, "gcsservices.careers.microsoft.com") {
+		u, err := url.Parse(targetURL)
+		if err == nil && u.Host != "" {
+			q := u.Query()
+			searchAPI = fmt.Sprintf("https://gcsservices.careers.microsoft.com/search/api/v1/search?%s", q.Encode())
+		} else {
+			searchAPI = fmt.Sprintf("https://gcsservices.careers.microsoft.com/search/api/v1/search?q=%s&lc=Redmond%%2C+Washington%%2C+United+States&l=en_us", url.QueryEscape(targetURL))
+		}
+	}
+
+	pageSize := 50
+	for page := 1; page <= 5; page++ {
+		reqURL := searchAPI
+		if strings.Contains(reqURL, "?") {
+			reqURL = fmt.Sprintf("%s&pg=%d&pgSz=%d", searchAPI, page, pageSize)
+		} else {
+			reqURL = fmt.Sprintf("%s?pg=%d&pgSz=%d", searchAPI, page, pageSize)
+		}
+
+		body, err := c.fetchURL(ctx, reqURL)
+		if err != nil {
+			if page == 1 {
+				return nil, err
+			}
+			break
+		}
+
+		var resp msSearchResponse
+		if err := json.Unmarshal([]byte(body), &resp); err != nil {
+			if page == 1 {
+				return nil, fmt.Errorf("failed to parse Microsoft search response: %w", err)
+			}
+			break
+		}
+
+		jobs := resp.OperationResult.Result.Jobs
+		if len(jobs) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, job := range jobs {
+			if job.JobID == "" {
+				continue
+			}
+
+			applyURL := fmt.Sprintf("https://jobs.careers.microsoft.com/global/en/share/%s", job.JobID)
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("microsoft:%s", job.JobID),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(jobs) < pageSize {
+			break
+		}
+	}
+
+	logger.Log.Infow("Microsoft discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *MicrosoftCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseMicrosoftJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	detailURL := fmt.Sprintf("https://gcsservices.careers.microsoft.com/search/api/v1/job/%s?lang=en_us", jobID)
+	body, err := c.fetchURL(ctx, detailURL)
+	if err == nil && json.Valid([]byte(body)) {
+		var detailResp msJobDetailResponse
+		if err := json.Unmarshal([]byte(body), &detailResp); err == nil && detailResp.OperationResult.Result.Title != "" {
+			j := detailResp.OperationResult.Result
+
+			var sb strings.Builder
+			if j.Description != "" {
+				sb.WriteString("## Overview\n\n")
+				sb.WriteString(cleanHTML(j.Description))
+				sb.WriteString("\n\n")
+			}
+			if j.Responsibilities != "" {
+				sb.WriteString("## Responsibilities\n\n")
+				sb.WriteString(cleanHTML(j.Responsibilities))
+				sb.WriteString("\n\n")
+			}
+			if j.Qualifications != "" {
+				sb.WriteString("## Qualifications\n\n")
+				sb.WriteString(cleanHTML(j.Qualifications))
+				sb.WriteString("\n\n")
+			}
+
+			loc := j.Properties.PrimaryLocation
+			if loc == "" && len(j.Properties.Locations) > 0 {
+				loc = strings.Join(j.Properties.Locations, "; ")
+			}
+			if loc == "" {
+				loc = "Redmond, WA, USA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("microsoft:%s", jobID),
+				Title:       j.Title,
+				Company:     "Microsoft",
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  sb.String(),
+			}, nil
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = "Software Engineer - Microsoft"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("microsoft:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Microsoft",
+		Location:    "Redmond, WA, USA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *MicrosoftCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseMicrosoftJobID(u string) string {
+	re := regexp.MustCompile(`/(?:share|job)/([0-9A-Za-z]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/microsoft/microsoft_test.go

@@ -0,0 +1,121 @@
+package microsoft
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestMicrosoftCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/job/1789001") {
+				body := `{
+					"operationResult": {
+						"result": {
+							"jobId": "1789001",
+							"title": "Software Engineer II - Azure Core",
+							"description": "<p>Azure Core is building the backbone of cloud compute.</p>",
+							"responsibilities": "<li>Design and implement scalable distributed services.</li>",
+							"qualifications": "<li>Bachelor degree and 2+ years of software development experience.</li>",
+							"properties": {
+								"primaryLocation": "Redmond, WA, United States"
+							}
+						}
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/search") || strings.Contains(r.URL.RawQuery, "pgSz") {
+				body := `{
+					"operationResult": {
+						"result": {
+							"totalJobs": 1,
+							"jobs": [
+								{
+									"jobId": "1789001",
+									"title": "Software Engineer II - Azure Core",
+									"properties": {
+										"primaryLocation": "Redmond, WA, United States",
+										"discipline": "Software Engineering"
+									}
+								}
+							]
+						}
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewMicrosoftCrawlerWithClient(mockClient)
+	if !c.CanHandle("microsoft") || !c.CanHandle("azure") {
+		t.Errorf("CanHandle should support microsoft and azure")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://gcsservices.careers.microsoft.com/search/api/v1/search?q=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "microsoft:1789001" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://jobs.careers.microsoft.com/global/en/share/1789001")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer II - Azure Core" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Microsoft" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Redmond, WA, United States" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestMicrosoftCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Microsoft",
+		Crawler:                 NewMicrosoftCrawler(),
+		TargetURL:               "https://gcsservices.careers.microsoft.com/search/api/v1/search?q=Software+Engineer&lc=Redmond%2C+Washington%2C+United+States&l=en_us",
+		ExpectedCanonicalPrefix: "microsoft:",
+		ExpectedCompany:         "Microsoft",
+	})
+}
+

pkg/crawler/netflix/netflix.go

@@ -0,0 +1,220 @@
+package netflix
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type NetflixCrawler struct {
+	client *http.Client
+}
+
+func NewNetflixCrawler() *NetflixCrawler {
+	return NewNetflixCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewNetflixCrawlerWithClient(client *http.Client) *NetflixCrawler {
+	return &NetflixCrawler{
+		client: client,
+	}
+}
+
+func (c *NetflixCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "netflix" || strings.Contains(p, "jobs.netflix.com")
+}
+
+type netflixSearchResponse struct {
+	Records struct {
+		Total int               `json:"total"`
+		Postings []netflixJobItem `json:"postings"`
+	} `json:"records"`
+}
+
+type netflixJobItem struct {
+	ID          int64    `json:"id"`
+	ExternalID  string   `json:"external_id"`
+	Title       string   `json:"text"`
+	Description string   `json:"description"`
+	Location    string   `json:"location"`
+	Locations   []string `json:"locations"`
+	Team        string   `json:"team"`
+	Department  string   `json:"department"`
+}
+
+func (c *NetflixCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	searchAPI := targetURL
+	if !strings.Contains(searchAPI, "api/search") {
+		u, err := url.Parse(targetURL)
+		if err == nil && u.Host != "" {
+			q := u.Query()
+			searchAPI = fmt.Sprintf("https://jobs.netflix.com/api/search?%s", q.Encode())
+		} else {
+			searchAPI = fmt.Sprintf("https://jobs.netflix.com/api/search?q=%s", url.QueryEscape(targetURL))
+		}
+	}
+
+	body, err := c.fetchURL(ctx, searchAPI)
+	if err != nil {
+		return nil, err
+	}
+
+	var resp netflixSearchResponse
+	if err := json.Unmarshal([]byte(body), &resp); err != nil {
+		return nil, fmt.Errorf("failed to parse Netflix search response: %w", err)
+	}
+
+	for _, job := range resp.Records.Postings {
+		jobID := job.ExternalID
+		if jobID == "" && job.ID != 0 {
+			jobID = fmt.Sprintf("%d", job.ID)
+		}
+		if jobID == "" {
+			continue
+		}
+
+		applyURL := fmt.Sprintf("https://jobs.netflix.com/jobs/%s", jobID)
+		if !seen[applyURL] {
+			seen[applyURL] = true
+			discovered = append(discovered, crawler.DiscoveredJob{
+				CanonicalID: fmt.Sprintf("netflix:%s", jobID),
+				ApplyURL:    applyURL,
+			})
+		}
+	}
+
+	logger.Log.Infow("Netflix discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *NetflixCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseNetflixJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	// Try API endpoint: https://jobs.netflix.com/api/search?q={id} or direct detail
+	detailAPI := fmt.Sprintf("https://jobs.netflix.com/api/postings/%s", jobID)
+	body, err := c.fetchURL(ctx, detailAPI)
+	if err == nil && json.Valid([]byte(body)) {
+		var job netflixJobItem
+		if err := json.Unmarshal([]byte(body), &job); err == nil && job.Title != "" {
+			loc := job.Location
+			if loc == "" && len(job.Locations) > 0 {
+				loc = strings.Join(job.Locations, "; ")
+			}
+			if loc == "" {
+				loc = "Seattle / Remote, WA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("netflix:%s", jobID),
+				Title:       job.Title,
+				Company:     "Netflix",
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  cleanHTML(job.Description),
+			}, nil
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = "Software Engineer - Netflix"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("netflix:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Netflix",
+		Location:    "Seattle / Remote, WA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *NetflixCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseNetflixJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/([0-9A-Za-z\-]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/netflix/netflix_test.go

@@ -0,0 +1,110 @@
+package netflix
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestNetflixCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/api/search") {
+				body := `{
+					"records": {
+						"total": 1,
+						"postings": [
+							{
+								"id": 8192301,
+								"external_id": "8192301",
+								"text": "Senior Software Engineer - Cloud Platform",
+								"location": "Seattle, WA, United States"
+							}
+						]
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "/postings/8192301") {
+				body := `{
+					"id": 8192301,
+					"external_id": "8192301",
+					"text": "Senior Software Engineer - Cloud Platform",
+					"description": "<p>We are looking for senior engineers to evolve Netflix telemetry and control planes.</p>",
+					"location": "Seattle, WA, United States"
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewNetflixCrawlerWithClient(mockClient)
+	if !c.CanHandle("netflix") {
+		t.Errorf("CanHandle should support netflix")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://jobs.netflix.com/api/search?q=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "netflix:8192301" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://jobs.netflix.com/jobs/8192301")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Senior Software Engineer - Cloud Platform" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Netflix" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Seattle, WA, United States" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestNetflixCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Netflix",
+		Crawler:                 NewNetflixCrawler(),
+		TargetURL:               "https://jobs.netflix.com/api/search?q=Software+Engineer",
+		ExpectedCanonicalPrefix: "netflix:",
+		ExpectedCompany:         "Netflix",
+	})
+}
+

pkg/crawler/nvidia/nvidia.go

@@ -0,0 +1,253 @@
+package nvidia
+
+import (
+	"bytes"
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type NvidiaCrawler struct {
+	client *http.Client
+}
+
+func NewNvidiaCrawler() *NvidiaCrawler {
+	return NewNvidiaCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewNvidiaCrawlerWithClient(client *http.Client) *NvidiaCrawler {
+	return &NvidiaCrawler{
+		client: client,
+	}
+}
+
+func (c *NvidiaCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "nvidia" || strings.Contains(p, "nvidia.wd5.myworkdayjobs.com")
+}
+
+type nvidiaSearchPayload struct {
+	AppliedFacets map[string][]string `json:"appliedFacets"`
+	Limit         int                 `json:"limit"`
+	Offset        int                 `json:"offset"`
+	SearchText    string              `json:"searchText"`
+}
+
+type nvidiaSearchResponse struct {
+	Total    int             `json:"total"`
+	JobPosts []nvidiaJobItem `json:"jobPostings"`
+}
+
+type nvidiaJobItem struct {
+	Title            string   `json:"title"`
+	ExternalPath     string   `json:"externalPath"`
+	BulletFields     []string `json:"bulletFields"`
+	LocationsText    string   `json:"locationsText"`
+	PostedOn         string   `json:"postedOn"`
+	JobPostingID     string   `json:"jobPostingId"`
+	TimeType         string   `json:"timeType"`
+}
+
+type nvidiaDetailResponse struct {
+	JobPostingInfo struct {
+		ID                   string   `json:"id"`
+		Title                string   `json:"title"`
+		JobDescription       string   `json:"jobDescription"`
+		Location             string   `json:"location"`
+		AdditionalLocations  []string `json:"additionalLocations"`
+		TimeType             string   `json:"timeType"`
+		StartDate            string   `json:"startDate"`
+		JobReqID             string   `json:"jobReqId"`
+	} `json:"jobPostingInfo"`
+}
+
+func (c *NvidiaCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	cxsURL := "https://nvidia.wd5.myworkdayjobs.com/wday/cxs/nvidia/NVIDIAExternalCareerSite/jobs"
+
+	searchText := "Software Engineer"
+	u, err := url.Parse(targetURL)
+	if err == nil && u.Query().Get("q") != "" {
+		searchText = u.Query().Get("q")
+	}
+
+	pageSize := 20
+	for page := 0; page < 5; page++ {
+		offset := page * pageSize
+		reqPayload := nvidiaSearchPayload{
+			AppliedFacets: make(map[string][]string),
+			Limit:         pageSize,
+			Offset:        offset,
+			SearchText:    searchText,
+		}
+
+		payloadBytes, err := json.Marshal(reqPayload)
+		if err != nil {
+			return nil, err
+		}
+
+		req, err := http.NewRequestWithContext(ctx, "POST", cxsURL, bytes.NewBuffer(payloadBytes))
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+
+		req.Header.Set("Content-Type", "application/json")
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+		req.Header.Set("Accept", "application/json")
+
+		resp, err := c.client.Do(req)
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+		defer func() { _ = resp.Body.Close() }()
+
+		body, err := io.ReadAll(resp.Body)
+		if err != nil {
+			break
+		}
+
+		var searchResp nvidiaSearchResponse
+		if err := json.Unmarshal(body, &searchResp); err != nil {
+			if page == 0 {
+				return nil, fmt.Errorf("failed to parse NVIDIA search response: %w", err)
+			}
+			break
+		}
+
+		if len(searchResp.JobPosts) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, post := range searchResp.JobPosts {
+			path := post.ExternalPath
+			if path == "" {
+				continue
+			}
+
+			jobID := parseNvidiaJobID(path)
+			if jobID == "" {
+				jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(path)))[:16]
+			}
+
+			applyURL := "https://nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAExternalCareerSite" + path
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("nvidia:%s", jobID),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(searchResp.JobPosts) < pageSize {
+			break
+		}
+	}
+
+	logger.Log.Infow("NVIDIA discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *NvidiaCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseNvidiaJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	cxsDetailURL := fmt.Sprintf("https://nvidia.wd5.myworkdayjobs.com/wday/cxs/nvidia/NVIDIAExternalCareerSite/job/%s", jobID)
+	req, err := http.NewRequestWithContext(ctx, "GET", cxsDetailURL, nil)
+	if err == nil {
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+		req.Header.Set("Accept", "application/json")
+		resp, err := c.client.Do(req)
+		if err == nil {
+			defer func() { _ = resp.Body.Close() }()
+			body, _ := io.ReadAll(resp.Body)
+			var detailResp nvidiaDetailResponse
+			if err := json.Unmarshal(body, &detailResp); err == nil && detailResp.JobPostingInfo.Title != "" {
+				info := detailResp.JobPostingInfo
+				var locs []string
+				if info.Location != "" {
+					locs = append(locs, info.Location)
+				}
+				for _, addLoc := range info.AdditionalLocations {
+					if addLoc != "" {
+						locs = append(locs, addLoc)
+					}
+				}
+				loc := strings.Join(locs, "; ")
+				if loc == "" {
+					loc = "Redmond / Seattle, WA"
+				}
+
+				return &crawler.ExtractedJob{
+					CanonicalID: fmt.Sprintf("nvidia:%s", jobID),
+					Title:       info.Title,
+					Company:     "NVIDIA",
+					Location:    loc,
+					ApplyURL:    applyURL,
+					RawContent:  cleanHTML(info.JobDescription),
+				}, nil
+			}
+		}
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("nvidia:%s", jobID),
+		Title:       "Software Engineer - NVIDIA",
+		Company:     "NVIDIA",
+		Location:    "Redmond / Seattle, WA",
+		ApplyURL:    applyURL,
+		RawContent:  "See job posting at " + applyURL,
+	}, nil
+}
+
+func parseNvidiaJobID(u string) string {
+	re := regexp.MustCompile(`/job/([^/?#]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/nvidia/nvidia_test.go

@@ -0,0 +1,110 @@
+package nvidia
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestNvidiaCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/job/JR1982000") {
+				body := `{
+					"jobPostingInfo": {
+						"id": "JR1982000",
+						"title": "Senior Systems Software Engineer - GPU Infrastructure",
+						"jobDescription": "<p>NVIDIA is seeking senior systems software engineers to accelerate AI data centers.</p>",
+						"location": "Redmond, WA",
+						"additionalLocations": ["Seattle, WA"]
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/jobs") {
+				body := `{
+					"total": 1,
+					"jobPostings": [
+						{
+							"title": "Senior Systems Software Engineer - GPU Infrastructure",
+							"externalPath": "/job/JR1982000",
+							"locationsText": "Redmond, WA",
+							"jobPostingId": "JR1982000"
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewNvidiaCrawlerWithClient(mockClient)
+	if !c.CanHandle("nvidia") {
+		t.Errorf("CanHandle should support nvidia")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite?q=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "nvidia:JR1982000" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAExternalCareerSite/job/JR1982000")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Senior Systems Software Engineer - GPU Infrastructure" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "NVIDIA" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Redmond, WA; Seattle, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestNvidiaCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "NVIDIA",
+		Crawler:                 NewNvidiaCrawler(),
+		TargetURL:               "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite?q=Software+Engineer",
+		ExpectedCanonicalPrefix: "nvidia:",
+		ExpectedCompany:         "NVIDIA",
+	})
+}
+

pkg/crawler/snowflake/snowflake.go

@@ -0,0 +1,225 @@
+package snowflake
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type SnowflakeCrawler struct {
+	client *http.Client
+}
+
+func NewSnowflakeCrawler() *SnowflakeCrawler {
+	return NewSnowflakeCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewSnowflakeCrawlerWithClient(client *http.Client) *SnowflakeCrawler {
+	return &SnowflakeCrawler{
+		client: client,
+	}
+}
+
+func (c *SnowflakeCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "snowflake" || strings.Contains(p, "careers.snowflake.com") || strings.Contains(p, "boards.greenhouse.io/snowflake")
+}
+
+type snowflakeJobsResponse struct {
+	Jobs []snowflakeJob `json:"jobs"`
+}
+
+type snowflakeJob struct {
+	ID          int64  `json:"id"`
+	Title       string `json:"title"`
+	AbsoluteURL string `json:"absolute_url"`
+	Location    struct {
+		Name string `json:"name"`
+	} `json:"location"`
+	Offices []struct {
+		Name     string `json:"name"`
+		Location string `json:"location"`
+	} `json:"offices"`
+	Departments []struct {
+		Name string `json:"name"`
+	} `json:"departments"`
+	Content string `json:"content"`
+}
+
+func (c *SnowflakeCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	apiURL := "https://boards-api.greenhouse.io/v1/boards/snowflake/jobs"
+	if targetURL != "" && strings.Contains(targetURL, "boards-api.greenhouse.io") {
+		apiURL = targetURL
+	}
+
+	body, err := c.fetchURL(ctx, apiURL)
+	if err != nil {
+		return nil, err
+	}
+
+	var resp snowflakeJobsResponse
+	if err := json.Unmarshal([]byte(body), &resp); err != nil {
+		return nil, fmt.Errorf("failed to parse Snowflake Greenhouse API response: %w", err)
+	}
+
+	for _, job := range resp.Jobs {
+		if job.ID == 0 {
+			continue
+		}
+
+		jobIDStr := fmt.Sprintf("%d", job.ID)
+		applyURL := job.AbsoluteURL
+		if applyURL == "" {
+			applyURL = fmt.Sprintf("https://boards.greenhouse.io/snowflake/jobs/%s", jobIDStr)
+		}
+
+		if !seen[applyURL] {
+			seen[applyURL] = true
+			discovered = append(discovered, crawler.DiscoveredJob{
+				CanonicalID: fmt.Sprintf("snowflake:%s", jobIDStr),
+				ApplyURL:    applyURL,
+			})
+		}
+	}
+
+	logger.Log.Infow("Snowflake discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *SnowflakeCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseSnowflakeJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	detailAPI := fmt.Sprintf("https://boards-api.greenhouse.io/v1/boards/snowflake/jobs/%s", jobID)
+	body, err := c.fetchURL(ctx, detailAPI)
+	if err == nil && json.Valid([]byte(body)) {
+		var job snowflakeJob
+		if err := json.Unmarshal([]byte(body), &job); err == nil && job.Title != "" {
+			loc := job.Location.Name
+			if loc == "" && len(job.Offices) > 0 {
+				var locs []string
+				for _, o := range job.Offices {
+					if o.Name != "" {
+						locs = append(locs, o.Name)
+					}
+				}
+				loc = strings.Join(locs, "; ")
+			}
+			if loc == "" {
+				loc = "Bellevue / Seattle, WA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("snowflake:%s", jobID),
+				Title:       job.Title,
+				Company:     "Snowflake",
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  cleanHTML(job.Content),
+			}, nil
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*class="app-title"[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	}
+	if title == "" {
+		title = "Software Engineer - Snowflake"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("snowflake:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Snowflake",
+		Location:    "Bellevue / Seattle, WA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *SnowflakeCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseSnowflakeJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/(\d+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/snowflake/snowflake_test.go

@@ -0,0 +1,97 @@
+package snowflake
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestSnowflakeCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.HasSuffix(r.URL.Path, "/jobs/4192301") {
+				body := `{
+					"id": 4192301,
+					"title": "Principal Software Engineer - Core Database Engine",
+					"content": "<p>Snowflake is the Data Cloud.</p>",
+					"location": {
+						"name": "Bellevue, WA"
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/jobs") {
+				body := `{
+					"jobs": [
+						{
+							"id": 4192301,
+							"title": "Principal Software Engineer - Core Database Engine",
+							"absolute_url": "https://boards.greenhouse.io/snowflake/jobs/4192301",
+							"location": {
+								"name": "Bellevue, WA"
+							}
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewSnowflakeCrawlerWithClient(mockClient)
+	if !c.CanHandle("snowflake") {
+		t.Errorf("CanHandle should support snowflake")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://boards-api.greenhouse.io/v1/boards/snowflake/jobs")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "snowflake:4192301" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://boards.greenhouse.io/snowflake/jobs/4192301")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Principal Software Engineer - Core Database Engine" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Snowflake" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Bellevue, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}

pkg/crawler/stripe/stripe.go

@@ -0,0 +1,226 @@
+package stripe
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type StripeCrawler struct {
+	client *http.Client
+}
+
+func NewStripeCrawler() *StripeCrawler {
+	return NewStripeCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewStripeCrawlerWithClient(client *http.Client) *StripeCrawler {
+	return &StripeCrawler{
+		client: client,
+	}
+}
+
+func (c *StripeCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "stripe" || strings.Contains(p, "stripe.com/jobs") || strings.Contains(p, "boards.greenhouse.io/stripe")
+}
+
+type greenhouseJobsResponse struct {
+	Jobs []greenhouseJob `json:"jobs"`
+}
+
+type greenhouseJob struct {
+	ID          int64  `json:"id"`
+	Title       string `json:"title"`
+	AbsoluteURL string `json:"absolute_url"`
+	Location    struct {
+		Name string `json:"name"`
+	} `json:"location"`
+	Offices []struct {
+		Name     string `json:"name"`
+		Location string `json:"location"`
+	} `json:"offices"`
+	Departments []struct {
+		Name string `json:"name"`
+	} `json:"departments"`
+	Content   string `json:"content"`
+	UpdatedAt string `json:"updated_at"`
+}
+
+func (c *StripeCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	apiURL := "https://boards-api.greenhouse.io/v1/boards/stripe/jobs"
+	if targetURL != "" && strings.Contains(targetURL, "boards-api.greenhouse.io") {
+		apiURL = targetURL
+	}
+
+	body, err := c.fetchURL(ctx, apiURL)
+	if err != nil {
+		return nil, err
+	}
+
+	var resp greenhouseJobsResponse
+	if err := json.Unmarshal([]byte(body), &resp); err != nil {
+		return nil, fmt.Errorf("failed to parse Stripe Greenhouse API response: %w", err)
+	}
+
+	for _, job := range resp.Jobs {
+		if job.ID == 0 {
+			continue
+		}
+
+		jobIDStr := fmt.Sprintf("%d", job.ID)
+		applyURL := job.AbsoluteURL
+		if applyURL == "" {
+			applyURL = fmt.Sprintf("https://boards.greenhouse.io/stripe/jobs/%s", jobIDStr)
+		}
+
+		if !seen[applyURL] {
+			seen[applyURL] = true
+			discovered = append(discovered, crawler.DiscoveredJob{
+				CanonicalID: fmt.Sprintf("stripe:%s", jobIDStr),
+				ApplyURL:    applyURL,
+			})
+		}
+	}
+
+	logger.Log.Infow("Stripe discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *StripeCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseStripeJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	detailAPI := fmt.Sprintf("https://boards-api.greenhouse.io/v1/boards/stripe/jobs/%s", jobID)
+	body, err := c.fetchURL(ctx, detailAPI)
+	if err == nil && json.Valid([]byte(body)) {
+		var job greenhouseJob
+		if err := json.Unmarshal([]byte(body), &job); err == nil && job.Title != "" {
+			loc := job.Location.Name
+			if loc == "" && len(job.Offices) > 0 {
+				var locs []string
+				for _, o := range job.Offices {
+					if o.Name != "" {
+						locs = append(locs, o.Name)
+					}
+				}
+				loc = strings.Join(locs, "; ")
+			}
+			if loc == "" {
+				loc = "Seattle / Remote, WA"
+			}
+
+			return &crawler.ExtractedJob{
+				CanonicalID: fmt.Sprintf("stripe:%s", jobID),
+				Title:       job.Title,
+				Company:     "Stripe",
+				Location:    loc,
+				ApplyURL:    applyURL,
+				RawContent:  cleanHTML(job.Content),
+			}, nil
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*class="app-title"[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	}
+	if title == "" {
+		title = "Software Engineer - Stripe"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("stripe:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Stripe",
+		Location:    "Seattle / Remote, WA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *StripeCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseStripeJobID(u string) string {
+	re := regexp.MustCompile(`/jobs/(\d+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/stripe/stripe_test.go

@@ -0,0 +1,98 @@
+package stripe
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestStripeCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.HasSuffix(r.URL.Path, "/jobs/6198201") {
+				body := `{
+					"id": 6198201,
+					"title": "Software Engineer - Infrastructure Platform",
+					"content": "<p>Stripe Infrastructure powers global financial networks.</p>",
+					"location": {
+						"name": "Seattle, WA"
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/jobs") {
+				body := `{
+					"jobs": [
+						{
+							"id": 6198201,
+							"title": "Software Engineer - Infrastructure Platform",
+							"absolute_url": "https://boards.greenhouse.io/stripe/jobs/6198201",
+							"location": {
+								"name": "Seattle, WA"
+							}
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewStripeCrawlerWithClient(mockClient)
+	if !c.CanHandle("stripe") {
+		t.Errorf("CanHandle should support stripe")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://boards-api.greenhouse.io/v1/boards/stripe/jobs")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "stripe:6198201" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://boards.greenhouse.io/stripe/jobs/6198201")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer - Infrastructure Platform" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Stripe" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Seattle, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}

pkg/crawler/uber/uber.go

@@ -0,0 +1,306 @@
+package uber
+
+import (
+	"bytes"
+	"context"
+	"crypto/sha256"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"regexp"
+	"strings"
+	"time"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler"
+	"git.pengzhan.dev/haopengzhan/hireme/internal/logger"
+)
+
+type UberCrawler struct {
+	client *http.Client
+}
+
+func NewUberCrawler() *UberCrawler {
+	return NewUberCrawlerWithClient(&http.Client{
+		Timeout: 30 * time.Second,
+	})
+}
+
+func NewUberCrawlerWithClient(client *http.Client) *UberCrawler {
+	return &UberCrawler{
+		client: client,
+	}
+}
+
+func (c *UberCrawler) CanHandle(platform string) bool {
+	p := strings.ToLower(platform)
+	return p == "uber" || strings.Contains(p, "uber.com/careers") || strings.Contains(p, "uber.com/global/en/careers")
+}
+
+type uberSearchPayload struct {
+	Limit  int                    `json:"limit"`
+	Page   int                    `json:"page"`
+	Params map[string]interface{} `json:"params"`
+}
+
+type uberSearchResponse struct {
+	Status string `json:"status"`
+	Data   struct {
+		TotalResults int           `json:"totalResults"`
+		Results      []uberJobItem `json:"results"`
+	} `json:"data"`
+}
+
+type uberJobItem struct {
+	ID                 int64    `json:"id"`
+	Title              string   `json:"title"`
+	Description        string   `json:"description"`
+	Location           string   `json:"location"`
+	City               string   `json:"city"`
+	Country            string   `json:"countryName"`
+	Department         string   `json:"department"`
+	Team               string   `json:"team"`
+	LineOfBusiness     string   `json:"lineOfBusiness"`
+	AllLocations       []string `json:"allLocations"`
+	CreationDate       string   `json:"creationDate"`
+	WhatYouWillDo      string   `json:"whatYouWillDo"`
+	BasicQualifications string   `json:"basicQualifications"`
+}
+
+func (c *UberCrawler) DiscoverJobs(ctx context.Context, targetURL string) ([]crawler.DiscoveredJob, error) {
+	seen := make(map[string]bool)
+	var discovered []crawler.DiscoveredJob
+
+	apiURL := "https://www.uber.com/api/loadSearchJobsResults"
+
+	keyword := "Software Engineer"
+	u, err := url.Parse(targetURL)
+	if err == nil && u.Query().Get("query") != "" {
+		keyword = u.Query().Get("query")
+	}
+
+	pageSize := 20
+	for page := 0; page < 5; page++ {
+		reqPayload := uberSearchPayload{
+			Limit: pageSize,
+			Page:  page,
+			Params: map[string]interface{}{
+				"query": keyword,
+			},
+		}
+
+		payloadBytes, err := json.Marshal(reqPayload)
+		if err != nil {
+			return nil, err
+		}
+
+		req, err := http.NewRequestWithContext(ctx, "POST", apiURL, bytes.NewBuffer(payloadBytes))
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+
+		req.Header.Set("Content-Type", "application/json")
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+		req.Header.Set("x-csrf-token", "x")
+
+		resp, err := c.client.Do(req)
+		if err != nil {
+			if page == 0 {
+				return nil, err
+			}
+			break
+		}
+		defer func() { _ = resp.Body.Close() }()
+
+		body, err := io.ReadAll(resp.Body)
+		if err != nil {
+			break
+		}
+
+		var searchResp uberSearchResponse
+		if err := json.Unmarshal(body, &searchResp); err != nil {
+			if page == 0 {
+				return nil, fmt.Errorf("failed to parse Uber search response: %w", err)
+			}
+			break
+		}
+
+		results := searchResp.Data.Results
+		if len(results) == 0 {
+			break
+		}
+
+		newFound := false
+		for _, job := range results {
+			if job.ID == 0 {
+				continue
+			}
+
+			jobIDStr := fmt.Sprintf("%d", job.ID)
+			applyURL := fmt.Sprintf("https://www.uber.com/global/en/careers/list/%s/", jobIDStr)
+
+			if !seen[applyURL] {
+				seen[applyURL] = true
+				newFound = true
+				discovered = append(discovered, crawler.DiscoveredJob{
+					CanonicalID: fmt.Sprintf("uber:%s", jobIDStr),
+					ApplyURL:    applyURL,
+				})
+			}
+		}
+
+		if !newFound || len(results) < pageSize {
+			break
+		}
+	}
+
+	logger.Log.Infow("Uber discovery finished", "found_count", len(discovered))
+	return discovered, nil
+}
+
+func (c *UberCrawler) FetchJobDetail(ctx context.Context, applyURL string) (*crawler.ExtractedJob, error) {
+	jobID := parseUberJobID(applyURL)
+	if jobID == "" {
+		jobID = fmt.Sprintf("%x", sha256.Sum256([]byte(applyURL)))[:16]
+	}
+
+	// Try API endpoint: https://www.uber.com/api/loadJobDetails?id={id}
+	detailAPI := fmt.Sprintf("https://www.uber.com/api/loadJobDetails?id=%s", jobID)
+	req, err := http.NewRequestWithContext(ctx, "GET", detailAPI, nil)
+	if err == nil {
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+		req.Header.Set("x-csrf-token", "x")
+		resp, err := c.client.Do(req)
+		if err == nil {
+			defer func() { _ = resp.Body.Close() }()
+			body, _ := io.ReadAll(resp.Body)
+			var detailResp struct {
+				Data uberJobItem `json:"data"`
+			}
+			if err := json.Unmarshal(body, &detailResp); err == nil && detailResp.Data.Title != "" {
+				j := detailResp.Data
+				var sb strings.Builder
+				if j.Description != "" {
+					sb.WriteString("## Description\n\n")
+					sb.WriteString(cleanHTML(j.Description))
+					sb.WriteString("\n\n")
+				}
+				if j.WhatYouWillDo != "" {
+					sb.WriteString("## What You'll Do\n\n")
+					sb.WriteString(cleanHTML(j.WhatYouWillDo))
+					sb.WriteString("\n\n")
+				}
+				if j.BasicQualifications != "" {
+					sb.WriteString("## Basic Qualifications\n\n")
+					sb.WriteString(cleanHTML(j.BasicQualifications))
+					sb.WriteString("\n\n")
+				}
+
+				loc := j.Location
+				if loc == "" && j.City != "" {
+					loc = fmt.Sprintf("%s, %s", j.City, j.Country)
+				}
+				if loc == "" {
+					loc = "Seattle, WA"
+				}
+
+				return &crawler.ExtractedJob{
+					CanonicalID: fmt.Sprintf("uber:%s", jobID),
+					Title:       j.Title,
+					Company:     "Uber",
+					Location:    loc,
+					ApplyURL:    applyURL,
+					RawContent:  sb.String(),
+				}, nil
+			}
+		}
+	}
+
+	// Fallback to HTML
+	htmlBody, err := c.fetchURL(ctx, applyURL)
+	if err != nil {
+		return nil, err
+	}
+
+	title := extractRegex(htmlBody, `<h1[^>]*>(.*?)</h1>`)
+	if title == "" {
+		title = "Software Engineer - Uber"
+	}
+
+	return &crawler.ExtractedJob{
+		CanonicalID: fmt.Sprintf("uber:%s", jobID),
+		Title:       cleanHTML(title),
+		Company:     "Uber",
+		Location:    "Seattle, WA",
+		ApplyURL:    applyURL,
+		RawContent:  cleanHTML(htmlBody),
+	}, nil
+}
+
+func (c *UberCrawler) fetchURL(ctx context.Context, urlStr string) (string, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
+	if err != nil {
+		return "", fmt.Errorf("failed to create request: %w", err)
+	}
+
+	req.Header.Set("User-Agent", "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
+	req.Header.Set("Accept", "application/json, text/html, */*")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("HTTP GET failed: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("HTTP GET returned status %d for %s", resp.StatusCode, urlStr)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("failed to read response: %w", err)
+	}
+
+	return string(body), nil
+}
+
+func parseUberJobID(u string) string {
+	re := regexp.MustCompile(`/list/([0-9A-Za-z]+)`)
+	m := re.FindStringSubmatch(u)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func extractRegex(s, pattern string) string {
+	re := regexp.MustCompile(pattern)
+	m := re.FindStringSubmatch(s)
+	if len(m) >= 2 {
+		return m[1]
+	}
+	return ""
+}
+
+func cleanHTML(s string) string {
+	s = strings.ReplaceAll(s, "<b>", "**")
+	s = strings.ReplaceAll(s, "</b>", "**")
+	s = strings.ReplaceAll(s, "<strong>", "**")
+	s = strings.ReplaceAll(s, "</strong>", "**")
+	s = strings.ReplaceAll(s, "<br>", "\n")
+	s = strings.ReplaceAll(s, "<br/>", "\n")
+	s = strings.ReplaceAll(s, "<br />", "\n")
+	s = strings.ReplaceAll(s, "<li>", "- ")
+	s = strings.ReplaceAll(s, "</li>", "\n")
+	s = strings.ReplaceAll(s, "<ul>", "\n")
+	s = strings.ReplaceAll(s, "</ul>", "\n")
+	s = strings.ReplaceAll(s, "<p>", "\n")
+	s = strings.ReplaceAll(s, "</p>", "\n")
+	reTags := regexp.MustCompile(`<[^>]*>`)
+	s = reTags.ReplaceAllString(s, "")
+	return strings.TrimSpace(s)
+}

pkg/crawler/uber/uber_test.go

@@ -0,0 +1,114 @@
+package uber
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"git.pengzhan.dev/haopengzhan/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestUberCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "loadSearchJobsResults") {
+				body := `{
+					"status": "success",
+					"data": {
+						"totalResults": 1,
+						"results": [
+							{
+								"id": 139820,
+								"title": "Software Engineer II - Core Compute & Storage",
+								"location": "Seattle, WA",
+								"city": "Seattle",
+								"countryName": "USA"
+							}
+						]
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "loadJobDetails") {
+				body := `{
+					"data": {
+						"id": 139820,
+						"title": "Software Engineer II - Core Compute & Storage",
+						"description": "<p>Uber Infra powers global dispatch and real-time computation.</p>",
+						"whatYouWillDo": "<li>Architect scalable storage solutions using Go and Java.</li>",
+						"basicQualifications": "<li>2+ years of backend software development experience.</li>",
+						"location": "Seattle, WA"
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewUberCrawlerWithClient(mockClient)
+	if !c.CanHandle("uber") {
+		t.Errorf("CanHandle should support uber")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://www.uber.com/global/en/careers/search/?query=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "uber:139820" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://www.uber.com/global/en/careers/list/139820/")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer II - Core Compute & Storage" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Uber" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Seattle, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestUberCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Uber",
+		Crawler:                 NewUberCrawler(),
+		TargetURL:               "https://www.uber.com/global/en/careers/search/?query=Software+Engineer",
+		ExpectedCanonicalPrefix: "uber:",
+		ExpectedCompany:         "Uber",
+	})
+}