Cairn

Cairn is an MCP server that turns any company's careers page into structured, queryable job data for AI agents. Point it at a Greenhouse board, a Lever listing, a Workday portal, or a plain HTML careers page — Cairn scrapes all postings without data loss, normalises them into a consistent schema, and exposes them as MCP tools that Claude (or any MCP-compatible agent) can call directly. No manual copy-pasting, no brittle spreadsheets: just clean job intelligence on demand.

---

Install

1. Clone and enter the repo

git clone https://github.com/chalshik/cairn.git
cd cairn

2. Create a virtual environment

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate

3. Install Python dependencies

pip install mcp httpx beautifulsoup4 playwright

4. Install the Playwright browser

playwright install chromium

---

Register with Claude Code

claude mcp add cairn -- python main.py

Verify it appears:

claude mcp list

---

MCP Tools

Tool	Description
scrape_jobs(url)	Scrape all job postings from a careers page. Returns a JSON array.
filter_jobs(jobs, keyword)	Filter the array by keyword (title / department / location / description).
get_job_detail(job_url)	Fetch the full description of a single posting by its URL.

Output schema (per job)

{
  "title":       "Senior Backend Engineer",
  "department":  "Engineering",
  "location":    "Remote – US",
  "url":         "https://boards.greenhouse.io/stripe/jobs/12345",
  "description": "Full plain-text description...",
  "posted_date": "2024-03-01T00:00:00Z"
}

---

Supported Platforms

Platform	Method
Greenhouse	Public API (boards-api.greenhouse.io) — fast and lossless
Lever	Public API (api.lever.co)
Workday / custom JS boards	Playwright (headless Chromium) with scroll + pagination
Static HTML pages	httpx + BeautifulSoup heuristic extraction

---

Example Prompts for Claude

Show me all open engineering roles at Stripe.
→ Use scrape_jobs("https://boards.greenhouse.io/stripe")

Find remote senior product manager roles at Notion.
→ Use scrape_jobs on Notion's careers page, then filter_jobs with "senior product manager remote"

What does Figma's Staff Designer role require?
→ Use get_job_detail("<job-url from scrape_jobs output>")

Compare backend engineering openings across Vercel and Linear.
→ scrape_jobs for each, then filter_jobs with "backend engineer" on both results

List all jobs in Berlin at Contentful.
→ scrape_jobs("https://www.contentful.com/careers/") then filter_jobs with "berlin"

---

Docker

Inside the container Cairn runs as an HTTP+SSE server on port 8000, so it stays alive and accepts connections like a real service.

Start with Docker Compose (recommended)

docker compose up --build

The server will be available at http://localhost:8000/sse.

Or build and run manually

docker build -t cairn .
docker run -p 8000:8000 cairn

Register with Claude Code (SSE transport)

claude mcp add --transport sse cairn http://localhost:8000/sse

Use stdio transport instead (no HTTP)

If you prefer the original stdio mode (process managed by Claude Code):

claude mcp add cairn -- docker run -i --rm -e CAIRN_TRANSPORT=stdio cairn

---

Project Structure

cairn/
├── main.py          # MCP server — tool registration and entry point
├── scraper.py       # Careers page scraper (Greenhouse API, Lever API, httpx, Playwright)
├── parser.py        # Normalisation, deduplication, grouping, and filtering
├── Dockerfile
├── requirements.txt
├── .dockerignore
└── README.md

---

Requirements

• Python 3.11+
• mcp — MCP Python SDK
• httpx — async-capable HTTP client
• beautifulsoup4 — HTML parsing
• playwright — headless browser for JS-rendered pages