luebangs-coder/Nemesis check
Run Nemesis detectors against a repository to catch agentic failure modes (e.g. an agent declaring success before verifying the real repo state).
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| repo | Path to the git repository to inspect. | no | . |
| transcript | Path to a file containing the agent transcript (optional). | no | "" |
| claimed-success | Set to 'true' if the agent declared the task complete. | no | false |
| tests-passing | Test outcome to record: 'true', 'false', or '' to omit. Nemesis never runs your tests itself; it only reasons about the outcome you provide. | no | "" |
| output | Optional path to write a Markdown report instead of stdout. | no | "" |
| fail-on-detect | Fail the action (non-zero exit) when any failure mode fires. | no | true |
| package-version | Version of the nemesis-eval package to install from PyPI (e.g. '0.2.0'). Leave empty to install the version bundled with this action ref. | no | "" |
| python-version | Python version to set up for the run. | no | 3.x |
Outputs
no outputs