badrshs/TestMedic - Self-Healing E2E Tests

Classifies Playwright failures, repairs UI drift in a verified PR, reports real bugs as issues.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
claude-code-oauth-tokenClaude subscription OAuth token (from `claude setup-token`). Provide this or anthropic-api-key.no""
anthropic-api-keyAnthropic API key. Provide this or claude-code-oauth-token.no""
modelModel used for diagnosis and healing.noclaude-sonnet-5
modeheal (fix and open a PR), dry-run (log proposals only), or issue-only (never patch, only report).noheal
max-healsMaximum number of tests to heal per run (cost cap and loop guard).no3
min-confidenceMinimum model confidence (0 to 1) required to apply a heal.no0.6
reportPath to the Playwright JSON report, relative to working-directory.notest-results/results.json
working-directoryDirectory containing the Playwright project, relative to the repository root.no.
test-commandCommand used to re-run a healed test for verification.nonpx playwright test
base-urlBase URL of the app under test, used to fetch live DOM evidence for healing. Defaults to the baseURL found in the Playwright config.no""
github-tokenToken used to push the heal branch and open PRs/issues.no${{ github.token }}
namedescription
pr-urlURL of the heal pull request, if one was opened.
issue-urlsComma-separated URLs of bug issues opened.
healed-countNumber of tests healed and verified.