cardscan-ai/Claude Code Watchdog
Your AI watchdog that watches for test failures and heals them automatically
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Maintainedlast commit Dec 20, 2025
- License
- MIT
Pinned Snippet
uses: cardscan-ai/claude-code-watchdog@00b5ba76f5f1e40a7d21f71f072149299bcb4a63 # v0.3.1tags can be moved; commit SHAs can't. why a SHA?
Inputs
| name | description | required | default |
|---|---|---|---|
| anthropic_api_key | Anthropic API key for Claude | yes | — |
| severity_threshold | Minimum severity to process (ignore|low|medium|high|critical) | no | medium |
| create_issues | Create GitHub issues for failures | no | true |
| create_fixes | Attempt to implement fixes automatically | no | true |
| rerun_tests | Attempt to re-run failed tests to confirm fixes | no | false |
| debug_mode | Upload watchdog artifacts for debugging | no | false |
| safe_mode | Skip potentially risky external content (GitHub issues, PRs, commit messages) | no | false |
| test_results_path | Path or glob pattern to test result files (e.g., "test-results/**/*.xml", "cypress/reports/*.json") | yes | — |
Outputs
| name | description |
|---|---|
| severity | Failure severity (ignore|low|medium|high|critical) |
| action_taken | What action was taken (issue_created|issue_updated|pr_created|pr_updated|tests_fixed|none) |
| issue_number | GitHub issue number if created or updated |
| pr_number | PR number if fixes were created |
| tests_passing | true if re-run tests passed after fixes |