🧭 Pathwise — Smart Career Navigation (Python MCP)

Pathwise is an intelligent career automation platform refactored to Python + FastMCP. Provides MCP tools for profile management, job search, cover letters, application tracking, and automated form filling.

🚀 Key Features

• Profile Management — Store and manage professional profiles with skills, experience, and education
• AI CV Parsing — Extract structured data from PDF CVs (regex or AI-powered via RouteMCP)
• Cover Letter Generation — Tailored cover letters using RouteMCP AI
• Job Search — Scrape 7 Chilean job boards (ChileTrabajos, CompuTrabajo, Laborum) + LinkedIn
• Smart Auto-Apply — AI-powered form detection and filling with BrowserMCP (Selenium)
• Application Tracking — Track applications with status, stats, search, and weekly reports
• Company Research — Research companies via Google + RouteMCP AI
• Interview Preparation — Generate questions and talking points from job descriptions

🛠️ Tech Stack

• Runtime: Python 3.11+
• Framework: FastMCP (MCP protocol via stdio)
• Database: SQLite (WAL mode) with FTS5 full-text search
• AI: RouteMCP (Llama 3.3 70B via Groq)
• Browser: Selenium + Chromium (external debug port 9226)
• PDF: PyMuPDF + PyPDF2
• Scraping: httpx + BeautifulSoup4
• ScrapeMCP: Optional generic fallback scraper

📦 MCP Tools (30 total)

Profile

• profile_load, profile_list, profile_save, profile_delete, profile_generate_personas

Job Search

• job_search, job_search_from_profile, linkedin_search, linkedin_scroll

Applications

• application_list, application_get, application_create, application_update, application_update_status, application_delete, application_stats, search_applications, get_weekly_report

Cover Letters

• cover_letter_generate, cover_letter_suggest_improvements

CV

• cv_parse_pdf

Setup

• scripts/setup.sh — Check Chrome/chromedriver, Python deps, and RouteMCP

Auto-Apply (BrowserMCP)

• auto_apply_pipeline, smart_analyze_form, smart_fill_form, fill_application_field, click_application_button, application_form_fields, batch_apply

Company Research

• company_research

Interview

• interview_prepare

🤖 MCP Connection Setup

Configure MCPs in Claude Code

MCPs are configured via ~/.mcp.json:

{
  "mcpServers": {
    "pathwise": {
      "command": "python",
      "args": ["/path/to/pathwise/server.py"]
    },
    "routemcp": {
      "command": "routemcp",
      "disabled": true,
      "note": "Configure ROUTEMCP_URL"
    },
    "scrapemcp": {
      "command": "scrapemcp",
      "disabled": true,
      "note": "Configure SCRAPEMCP_URL"
    },
    "browsermcp": {
      "command": "browsermcp",
      "disabled": true,
      "note": "Configure BROWSERMCP_URL"
    }
  }
}

Environment Variables

# Pathwise Configuration
PATHWISE_DB_PATH=pathwise.db
PATHWISE_PROFILE_PATH=profile.json

# External MCP URLs (when enabled)
ROUTEMCP_ENABLED=true
ROUTEMCP_URL=http://localhost:8000

SCRAPEMCP_ENABLED=true
SCRAPEMCP_URL=http://localhost:8001

BROWSERMCP_ENABLED=true
BROWSERMCP_URL=http://localhost:8002

🚥 Getting Started

# Check environment
bash scripts/setup.sh

# Install dependencies
uv sync

# Start Chrome with remote debugging (required for auto-apply)
DISPLAY=:0 chromium --no-sandbox --remote-debugging-port=9226 --user-data-dir=$HOME/snap/chromium/common/chromium &

# Run the MCP server
python server.py

🏛️ Project Structure

Pathwise/
├── server.py              # FastMCP entry point
├── config.py              # Environment configuration
├── pyproject.toml         # Python project config
├── AGENTS.md              # Agent instructions
├── database/
│   ├── __init__.py        # SQLite connection + schema seed
│   └── repos/
│       ├── profiles.py    # Profile CRUD
│       └── applications.py # Application CRUD
├── services/
│   ├── ai_provider.py     # RouteMCP AI integration
│   ├── cv_service.py      # PDF parsing
│   ├── job_service.py     # Job search + matching
│   ├── form_filler.py     # Smart form detection and field filling
│   ├── scraper_engine.py  # Unified retry + ScrapeMCP fallback
│   ├── company_research.py # Company research via Google + AI
│   └── scrapers/
│       ├── chiletrabajos.py
│       ├── computrabajo.py
│       ├── indeed.py
│       ├── getonboard.py
│       ├── trabajando.py
│       ├── remoteok.py
│       └── laborum.py        # JSON API with JWT auth
├── scripts/
│   └── setup.sh           # Environment setup checker
├── tools/                 # MCP tool definitions
│   ├── profile_tools.py
│   ├── job_tools.py
│   ├── application_tools.py
│   ├── auto_apply_tools.py
│   ├── cover_letter_tools.py
│   ├── cv_tools.py
│   └── interview_tools.py
├── web-dashboard/         # Frontend (Vanilla JS)
├── extension/             # Chrome Extension (MV3)
└── docs/                  # Documentation

---

Developed by MedalCode — Pathwise is your terminal for professional infiltration.