Vibeleaderboard
Index / tool
Visit github.com
Category
Developer Tools
Rank
No. 2941Tools index

Previous survey · No. 2712 ·

Pricing
Open Source
Type
TOOL
Use case
Software Testing & Quality
Interfaces
Agent Skill / Plugin
GitHub
1 stars
Date

About

Kvitansiya is a Claude Code Stop hook that checks an agent's claims such as pushed, deployed, or tests pass against the real world before the agent may stop. If a claim is false, the stop is blocked and the agent receives the evidence. In the README's example run, a deploy check found that production reported an older commit than the one pushed, and the agent then fixed its deploy script.

At a glance

Kvitansiya checks some completion claims in a Claude Code session before it ends. It is worth trying if you often need to verify that a reported push, deployment or test run happened. It checks supported claims against evidence such as remote Git refs, version endpoints and the session transcript.

Good fit

Claude Code users who want a second check on routine completion reports.

Consider something else when

Skip it if you need to verify database writes, sent messages or arbitrary wording.

Start here

Run the standalone check in a scratch project before installing the hook. Compare a claim about an existing nonempty file with one about a missing file. Inspect whether each result is verified, contradicted or uncertain.

Cost and requirements

MIT-licensed. Requires Python 3.9 or later and Git. Hook mode requires Claude Code. The checker does not require a model API key.

Limits to know

Pattern matching can miss unusual wording. Unavailable evidence can produce a warning rather than a block. The hook can block once per stop and stores results in a local log.

Related work

Tags

claude-codehookverificationdeploymentai-agentspython

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.