Get started
Use from your agent
How Claude Code, Codex and other agents learn to send finished work to land.
The land skill
The installer gives Claude Code and Codex the land skill. It teaches them the whole loop, including how to fix a refusal, when to wait and how to push. With the skill installed you can simply say "land this".
Repository rules
For any other agent, or to make land the rule in one repository, add these lines to CLAUDE.md (AGENTS.md for Codex and others).
When work is done, run `git fetch origin main`, then `land merge <branch>... --onto main --json`. land merges the branches with main and runs the tests on the result. It prints JSON: `landed`, and `refused` with the failing tests. Fix a refusal and run it again. Then run the `local.push` command it prints. It sends the result together with land's signature: `git push --atomic origin <commit>:refs/heads/main refs/land/attestations`. With several agents, each works in its own worktree and branch, and one agent runs `land merge` with every branch. Never edit, skip or delete a test to get work through.
Docs for agents
Agents that look things up on their own can read llms.txt, a short guide written for them, and the skill file itself.
Made for agents
- Every command takes
--json, andland mergeprints exactly one JSON object. - Every error says the exact command that fixes it.
- Running the same
land mergeagain is how an agent waits. It never starts the merge twice. - Exit codes tell an agent what to do next without reading the text. See exit codes.
Never put a land key in a repository or in an agent's instructions. Agents on your machine use your login, and agents in CI use a token, see CI and GitHub Actions.