northgardtracker/rootmark
Grounded verification for AGENTS.md and AI-agent instructions. Verifies that the Node/npm/pnpm commands written in AGENTS.md-style instruction files match the repository's actual package metadata, without executing anything, with an optional sticky PR summary comment.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| root | Root path to scan for instruction files. | no | . |
| fail-on | Severity threshold that fails the job: error, warning, or off. Defaults to off (report-only) so the Action never fails a job unless the user explicitly opts in. | no | off |
| comment | Post or update a sticky, top-level PR summary comment with findings. Only applies to pull_request events. | no | false |
| github-token | Token used to post the PR summary comment when comment is true. | no | ${{ github.token }} |
| node-version | Node.js version used to run the CLI. | no | 22 |
Outputs
| name | description |
|---|---|
| findings-count | Total number of findings reported by the scan. |
| json-path | Absolute path to the JSON results file written during the scan. |