RegressionRadar

About RegressionRadar

Flags tests that still pass but whose actual output quietly changed after an agent's diff.

Price: $19 one-time

Included free

• Dual-run comparison: pre-agent commit vs post-agent diff (Jest/Vitest)
• Silent-change flagging with before/after output

Unlocked with a license

• pytest support
• Risk-weighted ranking via the dependency-graph hotspot score

Getting a license

Open the RegressionRadar sidebar (activity bar icon) and click Get License. After purchase you'll receive a license key — paste it into the sidebar and click Activate License. Keys are machine-locked; use Move to another machine… in the sidebar before reinstalling on a new device.

How comparison works

The first run of Compare Pre/Post-Agent Test Output captures a baseline (.regressionradar/baseline.json) instead of trying to automatically stash/checkout a prior git ref, which would be risky to script safely. Every later run diffs the current test results against that saved baseline and offers to replace it. "Silent output change" detection is an honest approximation: it diffs the raw console/reporter text captured around each test file between runs, grouped by nearest file mention in the output — it is not a structured value diff, since Jest/Vitest JSON reporters don't expose returned/logged values. Hotspot ranking (licensed) maps a test file to a same-named source file in a sibling src/-shaped folder; when that mapping isn't confident, the file is left unranked rather than guessed.