andreagriffiths11/proof-agent-verify
Adversarial verification for AI-generated code using GitHub Models API or custom providers (Anthropic, Azure OpenAI, OpenAI-compatible)
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| github-token | GitHub token for authentication (defaults to GITHUB_TOKEN) | no | ${{ github.token }} |
| models-token | GitHub token with models:read permission (required for GitHub Models API access). If not provided, falls back to github-token. | no | "" |
| base-ref | Git ref to compare against (e.g., origin/main, HEAD~1) | no | origin/main |
| force | Force verification even if threshold not met (bypasses file count checks) | no | false |
| block-on-fail | Exit with code 1 if verification fails (blocks PR merge) | no | true |
| post-comment | Post verification result as PR comment | no | true |
| comment-mode | Comment format: full, summary, or collapse (default: collapse) | no | collapse |
| max-comment-length | Maximum comment length in characters (default: 2000) | no | 2000 |
Outputs
| name | description |
|---|---|
| verdict | Verification verdict (PASS, FAIL, or PARTIAL) |
| summary | Verification summary (full text) |