Keel is the governance layer for agent-driven engineering. It traces every change from requirement to passing test, lets deterministic gates decide what merges, and keeps people in charge of direction.
Code that compiles and passes its own tests — but does something nobody asked for.
An AI reviewer approving AI code is an opinion, not a verdict. Auditors won't accept it, and neither should you.
Six months later, nobody can say why a line exists or which requirement it serves.
Requirement → acceptance criteria → test scenario → test case → unit test → review → merge. Each link is a record with a stable ID, versioned and queryable.
Readiness and merge are decided by checks computed in code. Tests are frozen before coding starts; an agent can't edit the test to fit its code.
Direction, expected test values and governance changes need a human approval the server enforces — agents cannot approve them, by design.
| People decide | Agents do |
|---|---|
| What to build, and in what order | Draft and evaluate requirements |
| What each test must expect | Design test scenarios and cases |
| Approval of rules and architecture decisions | Write the code and the unit tests |
| Credentials and access | Review, merge through gates, release |
Keel is developed under its own governance. DynoInvoice, an invoicing system built with Keel, is open on GitHub — every commit traceable on dynokore.dev.
We're opening Keel to a small number of engineering teams already building with agents.