Get back to your main branch faster
land merges each finished branch with main and reruns only the test commands that change can affect.
curl -fsSL https://useland.dev/install | sh land login
- Claude Code
- Codex
- any agent that runs a command
An answer your agent can act on
One lands. The other comes back with the failing test, the assertion and a command that reproduces it.
- Keeps your tests intact
A change that edits, skips or deletes an existing test is refused before anything runs.
- No person in the loop
The verdict is JSON. The agent fixes the code and runs
land mergeagain.
$ land merge sub/api sub/ledger --onto main{ "landed": ["sub/api"], "refused": [ { "who": "sub/ledger", "why": "tests_failed", "failing": [ { "test": "tests/test_ledger.py::test_reverse", "assert": "assert 110 == 90" } ], "repro": { "commit": "071366837c", "run": "python -m pytest tests/test_ledger.py…" }, "hitBy": "sub/api" } ], "guarding": "3/3", "tests_changed": [], "local": { "push": "git push --atomic origin 8364fb3…" }}Merges by entity, not by line
Two agents editing different entities (functions, classes, methods) in the same file are combined automatically, then tested together on top of your latest main.
def parse(s): return s.strip().split(",") <<<<<<< maindef slugify(s): return s.replace(" ", "-")=======def truncate(s, n): return s[:n]>>>>>>> agent-b
def parse(s): return s.strip().split(",") def slugify(s):agent a return s.replace(" ", "-") def truncate(s, n):agent b return s[:n]
Reruns only the test commands a change can reach
land records every file each test command reads. A command reruns only when one of those files changed, and the rest keep their recorded result.
What we measured, pass or fail
4 agent changes and 1 bug reaching main, land against a pull request flow with 11 CI runs
when main moves, run the same merge again: tests land already vouched for are carried over, not rerun, and there is no second CI run
in production, a check that failed and then passed on the same code was named flaky and the change was refused
For each number we wrote down the pass criteria before measuring, and we report the result whether it passed or failed. Where land falls short is under Limits.
Set up in four steps
- Install the land CLI and sign in
On the machine your agents run on. It signs you in with GitHub.
- Start your free day
No card. Pick a plan from your dashboard to keep landing after it.
- Add the repository, once
land finds the tests and records which ones already fail.
- Tell your agents where finished work goes
A few lines in CLAUDE.md or AGENTS.md. The installer adds the land skill too.
# 1 · install and sign in curl -fsSL https://useland.dev/install | sh land login # 3 · add the repository land add # when work is done land merge <branch>... --onto main
Priced on results, not minutes
You pay only for changes land puts on main. Refused changes are free, and so is your first day.
Both plans start with a 1-day free trial after checkout. You can cancel any time from your dashboard.