Installation¶

At the end you will have your preferred way to run agent-job-matcher: a local pip install, the published Docker image, or the fully containerized demo stack. Every path starts with the same .env bootstrap:

git clone https://github.com/senthilsweb/agent-job-matcher && cd agent-job-matcher
cp .env.example .env      # then fill in the values you need

No configuration or credential ever ships in source — a missing required value is a clear startup error, on purpose (AGENTS.md rule 5).

Option A — pip (CLI, Python package, local dev)¶

Python 3.11+:

pip install -e "backend[dev]"
jobmatch --help
pytest backend -m "not live"    # offline suite proves the install, no API key needed

This gives you the jobmatch CLI and the embeddable core (from job_matcher import run_analysis, score_job_fit, extract_jsonresume). Add the [otel] extra if you plan to use an OTLP-shaped telemetry backend (Configuration).

Option B — the Docker image¶

Published to GHCR by CI on every push to main and on version tags:

docker pull ghcr.io/senthilsweb/agent-job-matcher:latest
docker run --rm --env-file .env ghcr.io/senthilsweb/agent-job-matcher:latest --help

Or through compose, backend only:

docker compose up -d api                 # FastAPI on :6010
docker compose run --rm cli version      # the CLI through the same image

Option C — the full demo stack¶

Every piece containerized, one consistent 6010-series port each:

docker compose up -d
# api            http://localhost:6010  — REST backend
# agent-service  http://localhost:6011  — chat bridge (LLM-2 + the MCP server inside)
# playground     http://localhost:6012  — resume + job links -> visual fit reports
# openapi-docs   http://localhost:6013  — browsable, branded API reference
# chat-demo      http://localhost:6014  — the mcp-chat-client chatbot demo page

The chat demo uses the mcp-chat-client widget — a runtime dependency, already baked into its container.

Claude Desktop (MCP)¶

The MCP server (mcp/) is a pure stdio bridge to the REST API — point Claude Desktop at it and the backend must be reachable. Setup and the tool list: mcp/README.md.

Next: Configuration.