agent-axiom/Agent Anvil
Run Agent Anvil scenario suites in GitHub Actions.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| scenario | Path to the Agent Anvil YAML scenario suite. | yes | — |
| runs-dir | Directory where run artifacts should be written. | no | runs/agent-anvil |
| offline | Use local heuristic semantic grading instead of OpenAI grading. | no | false |
| agent-mode | Demo agent mode, such as offline, openai, or auto. | no | "" |
| trials | Optional trial-count override. | no | "" |
| expected-exit-code | Expected anvil run exit code. Use 1 for intentional regression demos. | no | 0 |
| github-summary | Append a Markdown run summary to GITHUB_STEP_SUMMARY. | no | true |
| pr-comment | Write PR-ready Markdown for optional gh pr comment publishing. | no | false |
| post-pr-comment | Publish the generated PR comment with gh pr comment during pull_request workflows. | no | false |
| pr-comment-path | Path for the generated PR comment Markdown file. | no | agent-anvil-pr-comment.md |
| compare-baseline | Optional baseline run directory for compare-aware PR comments. | no | "" |
| compare-path | Path for the generated compare JSON artifact. | no | agent-anvil-compare.json |
| python-version | Python version used by uv for running Agent Anvil. | no | 3.12 |
| uv-cache | Enable setup-uv caching. Keep false when invoking this action multiple times in one job. | no | false |
Outputs
no outputs