MARTINLOOP

MartinLoop — Give coding agents more work. Watch them less. Ship more.

MartinLoop keeps coding-agent jobs focused, bounded, checked, recoverable, and accountable so you can delegate real software work without supervising every step.

Open source · Apache 2.0 licensed · martin-loop on npm · Keesan12/martin-loop on GitHub

npx -y martin-loop@latest start
# Alternative: install globally
npm install -g martin-loop@latest
martin doctor --engine claude --cwd .
martin estimate "fix the failing test" --engine claude --cwd . --budget-usd 3 --max-iterations 2 --verify "npm test"
martin session-start --host claude --cwd .
martin preflight "fix the failing test" --engine claude --cwd . --budget-usd 3 --max-iterations 2 --verify "npm test"
martin run "fix the failing test" --engine claude --cwd . --budget-usd 3 --max-iterations 2 --verify "npm test"
martin dossier --latest
martin runs verify --latest

Give coding agents more work. Watch them less. Ship more.

MartinLoop gives supported coding-agent work a clear job contract: objective, scope, limits, checks, recovery, outcome, and run history. Runs close as VERIFIED, STOPPED, or NEEDS REVIEW. VERIFIED means the configured checks passed; it is not a universal correctness or security guarantee.

Not another agent. Not another IDE.

The coding agent still writes the code. MartinLoop keeps the job focused, bounded, checked, recoverable, and recorded. Native coding-agent CLIs and raw model endpoints are different execution paths; only publicly released and validated paths are described as full coding-worker support.

How does MartinLoop work?

  1. Define the job. Set the objective, scope, limits, and checks before work starts.
  2. Let the supported coding agent work. The agent edits the repository while MartinLoop keeps the job inside the configured contract.
  3. Stop when a boundary fires. Spend, attempt, policy, and no-progress limits can end a run instead of letting it continue indefinitely.
  4. Check the result. A run is VERIFIED only when the configured verification evidence passes.
  5. Keep the record. Run history, stop reasons, failure classes, recovery evidence, and receipt integrity remain inspectable afterward.

FAQ

What is MartinLoop?
The system around the coding agents you already use. Every run gets a budget, stop rule, verifier, rollback path, and run record — without changing which agent or model you use.
Is this another coding agent?
No. Claude, Codex, and Gemini are the workers. MartinLoop runs the job around the work they do, right in your CLI or IDE. Nothing to migrate to.
Is MartinLoop open source?
Yes. The core is Apache 2.0 licensed. The hosted dashboard and managed team features are commercial.
Who is it for?
Engineering teams, platform teams, and CTOs running AI coding agents in production who need control, auditability, and a record of what happened.