Developers
The loop you actually work in.
Create an agent, run its tests, deploy it to an environment, tail its logs, roll it back. Built for the loop a developer spends the day in rather than as a thin wrapper somebody wrote once.
Local runs are real runs
An agent run locally still authenticates as itself, still passes the permission checks and still produces an audit record marked as local. Not a mock that behaves differently from production.
Logs stream, with the tool calls in them
Tail a running agent and see its tool calls, waits and decisions as they happen. Not a summary after the fact.
Rollback is one command
Return an agent to a previous version, with its permissions, boundaries and approval rules. The permissions come back too — rolling back logic and leaving access widened would be the worst outcome.
It is honest about what needs a person
A production deploy prints what needs approving and who can approve it, then waits. It does not pretend to have deployed something that is queued for review.
Roughly what it looks like
Illustrative and deliberately short. The reference is at docs.agentlabora.com, and pretending a marketing page is documentation wastes an afternoon.
The working loop
bashlabora test --suite permissions
labora deploy --env staging
labora logs invoice-summary --follow
labora rollback invoice-summary --to v1.4Where it stops
The CLI cannot approve a production deploy, decide an approval, or grant a capability that is refused at the platform. It also has no offline mode that skips permission checks — an agent that behaves differently on your laptop than in production is an agent you cannot trust either place.
Developer platform
The rest of it.
Give AI a job.
Build an agent, choose what it can access, define what it may do, and deploy it to work safely across your business.
No code required · Full SDKs and API · Stop any agent instantly · [email protected]