xfurti/Leogriel behavioral regression
Run paired Agent Skill regressions and publish GitHub-native reports
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| skill | Installed project skill name or project-relative skill directory | yes | — |
| compare | Optional Git ref used as the immutable reference skill | no | "" |
| agent | AgentRunner ID | no | codex |
| runner-version | Exact Codex or Claude Code CLI version; latest is convenient but not reproducible | no | latest |
| model | Exact runner model ID; strongly recommended for comparable results | no | "" |
| runs | Sequential paired runs per test case (1-20) | no | 3 |
| working-directory | Project directory containing agent-skills.json | no | . |
| leogriel-version | npm version or dist-tag of @leogriel/cli | no | next |
| frozen-install | Restore project skills from agent-skills.lock before testing | no | true |
| trust-tests | Allow command assertions from the checked-out test YAML | no | false |
| artifact-name | Name of the uploaded regression-report artifact | no | leogriel-regression |
| comment-on-pr | Create or update the Leogriel pull-request comment | no | false |
| github-token | Token used only for an optional pull-request comment | no | "" |
Outputs
| name | description |
|---|---|
| verdict | improved, unchanged, regressed, or inconclusive |
| result-json | Absolute path to the redacted JSON result |
| markdown-report | Absolute path to the Markdown report |
| html-report | Absolute path to the HTML report |
| badge-json | Absolute path to a Shields endpoint badge document |