resume-mcp

MCP server for maintaining a professional profile database and generating tailored, ATS-optimized resumes for specific job postings.

Features

• Profile database — SQLite store for contact info, education, work experience, projects, and skills with proficiency ratings

• Job scraping — fetch and parse job postings from URLs; extract tech stack, experience level, requirements, and salary automatically

• Tailored resume generation — scores every experience and project against the job, selects the most relevant ones, and produces a clean LaTeX resume

• Relevance scoring — deterministic match score based on tech overlap, keyword coverage, and description similarity

• ATS/keyword analysis — identifies keyword gaps, naming mismatches (React vs React.js), weak action verbs, bullets lacking metrics, and unused profile items that could strengthen the resume

• No extra API keys — all analysis is pure Python (regex + keyword matching). The calling LLM (Claude, GPT, etc.) does the creative rewriting using its own subscription

• Application tracker — auto-logs every generated resume to a formatted applications.xlsx spreadsheet with color-coded scores and status tracking

Related MCP server: cv-forge-mcp

How tailoring works

Each experience and project is scored 0–1 against the job posting:

Items are ranked by score, the top 4 experiences and top 4 projects are selected, and skills matching the JD are reordered to appear first.

Review pipeline

1. full_pipeline(url) — scrapes job → generates initial resume → runs deterministic analysis in one call

2. The analysis returns: keyword gaps, ATS issues (exact phrasing mismatches), weak verb flags, bullets missing metrics, and unused profile items from the DB

3. The calling LLM acts as a hiring manager and rewrites bullets, weaves in missing keywords, reorders sections

4. save_optimized_resume(...) — saves the LLM-rewritten LaTeX back to the DB and compiles to PDF

No API keys are consumed in step 2. The LLM you're already talking to does the creative work.

Setup

cd ~/resume-mcp
pip install -r requirements.txt

# seed_profile.py is a template — open it and replace the placeholder data
# with your own info, then run it to populate the database:
python seed_profile.py

Tip: To keep your personal data out of git, copy the template to a private file:

cp seed_profile.py seed_yourname.py
echo "seed_yourname.py" >> .gitignore
python seed_yourname.py

Optional: Install pdflatex to compile resumes to PDF locally. Without it, .tex files are saved and can be compiled via Overleaf.

Claude Desktop config

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "resume-builder": {
      "command": "python3",
      "args": ["/Users/your-username/resume-mcp/server.py"]
    }
  }
}

Data storage

MCP tools

Profile

Job scraping

Resume generation

Review & optimization

Application tracker

Usage examples

These are natural language commands you'd send to Claude with the MCP server connected:

Generate a resume for this job posting: [paste URL]

Scrape this job and run the full pipeline so you can optimize my resume: [URL]

Add a new experience: I worked at Acme Corp as a backend engineer from Jan 2024 to present.
Bullets: [...]

What's my match score for a job that requires React, TypeScript, and AWS experience?

Update my application status for row 3 to "Interview"

Show me my application stats