- Who it helps
- Developers who need an inspectable definition of a completed coding task.
- Availability
- The public Apache-2.0 repository documents Claude Code and Node.js 20 or newer. Checks must be configured; an empty setup verifies nothing.
Define the boundary
The Stop hook runs configured checks. The default permits stopping after three blocked attempts, so exhaustion must be treated as a failure in downstream review.
Read the receipt narrowly
A signature helps track an observed check run. The README explains that an agent sharing the same user can access signing material; the receipt is not proof against that adversary.
What you can do
- Write a small failing acceptance test and define how retry exhaustion will appear in review before considering integration. Open reference ↗
Where the evidence stops
README inspected only. Hooks, signing and tests were not executed.
Sources read. Primary text checked October 19, 2026 UTC. The external product or research workflow was not executed; reported results remain attributed to their authors.
Sources & further reading
GitHub · implementation
Krites repository and failure boundaries ↗
README read for configuration, protected tests, Stop hook, retry exhaustion and signed receipt threat model.
Krites maintainers · page checked 2026-10-19
Coverage note: The linked sources establish the documented scope. Unlisted social platforms and independent execution remain unverified.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
Krites adds protected tests and configured checks to a Claude Code workflow. It evaluates command results when the agent attempts to finish and records a signed receipt.