Tewip 0.34.0

Tewip proves software work — written by a person or by an agent. It triages a failing test to a cause with evidence, says whether a test could ever fail, breaks your code to find out whether the tests would notice, reads what a migration will do to a live database, and gives one verdict on whether a change is done. It answers "cannot say" rather than guessing.

Install

npm i -g tewip-ai

As a GitHub Action:

- uses: hmamut39/tewip@v0.34.0
  if: always()
  with:
    openai-api-key: ${{ secrets.OPENAI_API_KEY }}   # optional: without it only the rules run

Also available as a Claude Code plugin, in the MCP registry, and on Open VSX for Cursor, Windsurf and VSCodium.

What it does

Command	Question it answers
tewip explain	Why did each test fail? A real bug, flaky, the environment, test data, or a changed selector
tewip vouch	Could these tests ever fail at all?
tewip prove	Would the tests catch a bug in this file?
tewip pick	Which tests could notice this change?
tewip schema	What will this migration do to a live database?
tewip standards	Does this change keep the rules our team wrote down?
tewip done	Is this work actually finished?

What is measured

Every number comes from strangers' real code, never from fixtures:

• Failure triage: 86–99% accurate across six real labelled datasets, $1.12 per 1,000 failures
• vouch: 19 of 19 correct on 5,448 tests in 8 public projects
• prove: found real gaps in expressjs/express
• schema: 259 real Zulip and Mastodon migrations read
• pick: 0 of 10 catching tests lost while running 7–9% of the suite

What is not measured is said so plainly, in the README and at the bottom of every report.

Built by Mamuti Aishan (Hasan).