valani9/vstack agent-quality gate
Diagnose an AI-agent trace with 34 organizational-behavior patterns; fail the build on findings.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| trace | Path to the agent trace JSON to diagnose. | yes | — |
| fail-on | Minimum finding severity that fails the build, one of: none | trace | low | moderate | medium | high | critical. Use 'none' to report without ever failing. | no | high |
| mode | Analysis depth: quick | standard | forensic. | no | standard |
| recipe | Optional named recipe bundle (see `vstack-recipes`). | no | "" |
| client | LLM client for analyzers that need one: none | anthropic | openai | ollama. Defaults to 'none' (deterministic analyzers only, no API key needed). For full findings, set 'anthropic'/'openai' and provide the matching API key via env. | no | none |
| shape | Force trace shape: individual | team | org (default: inferred). | no | "" |
| sarif | If set to a path, also write a SARIF 2.1.0 report there (upload it with github/codeql-action/upload-sarif to surface findings in code scanning). | no | "" |
| comment | If set to a path, write the findings summary (Markdown) there so you can post it as a sticky PR comment (e.g. marocchino/sticky-pull-request-comment). | no | "" |
| version | valanistack version to install (e.g. '0.45.0'). Empty = latest. | no | "" |
| python-version | Python version to set up. | no | 3.13 |
Outputs
| name | description |
|---|---|
| max-severity | Highest finding severity seen (or 'none'). |
| findings-count | Total number of findings. |
| report | Path to the written JSON report. |
| sarif | Path to the written SARIF report (empty unless `sarif` input set). |
| comment | Path to the written Markdown comment (empty unless `comment` input set). |