karunyagoyal/FlakeHunt — AI Flaky Test Root-Cause Fixer
Classifies why your tests are flaky and suggests a code fix. Works with Anthropic, Gemini, Groq, GitHub Models, or rule-based (no API).
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| model-provider | AI provider to use for classification and fix generation. Options: anthropic — Claude Haiku (classify) + Claude Sonnet (fix) [paid, ~$0.01/run] gemini — Gemini 1.5 Flash [free tier available] groq — Llama 3.1 8B (classify) + Llama 70B (fix) [free tier available] github — GPT-4o-mini (classify) + GPT-4o (fix) [free with GitHub account] rules-only — keyword matching, no API call [always free, no fix generation] | no | anthropic |
| api-key | API key for the chosen model-provider. anthropic → Anthropic API key (console.anthropic.com) gemini → Google AI API key (aistudio.google.com) groq → Groq API key (console.groq.com) github → not needed — uses github-token automatically rules-only → not needed | no | — |
| anthropic-api-key | Deprecated — use api-key instead. Kept for backward compatibility. | no | — |
| github-token | GitHub token for reading logs and posting PR comments. Also used as the API key when model-provider is "github". | yes | ${{ github.token }} |
| confidence-threshold | Minimum confidence score (0-100) required to post a fix. Skips fix generation below this threshold to save API cost. | no | 70 |
Outputs
| name | description |
|---|---|
| root-cause | Classified root cause category (e.g. SELECTOR_FRAGILITY) |
| confidence | AI confidence score (0-100) |