justinkastil/PromptPolygraph
Run the PromptPolygraph evaluation gate (run -> analyze --ci -> report) and emit JUnit/SARIF.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| config | Path to the config.yaml under test. | yes | — |
| mode | Corpus mode passed to `polygraph all` (fixed, varied, adversarial, hybrid). Empty uses the config default. | no | "" |
| format | Comma list of report formats for the human-facing render. | no | md,html |
| baseline | Baseline for the regression check (run id, rolling:N, or HEAD). Empty disables it. | no | "" |
| mock | Offline deterministic mode (no API calls). Set "true" for key-free runs. | no | false |
| python-version | Python version for actions/setup-python. | no | 3.12 |
| extras | Optional pip extras to install with the package, e.g. "llm,pdf". Empty installs core only. | no | "" |
| source | Install source for the package. The PyPI name "promptpolygraph" by default, or "." to install the checked-out repo. | no | promptpolygraph |
Outputs
| name | description |
|---|---|
| verdict | Gate result, "pass" or "fail". |
| report-dir | Directory holding the run artifacts (reports, JUnit, SARIF, summary.json). |