aretedriver/Context Hygiene
Audit LLM conversation context window hygiene. Detects staleness, contradictions, deadweight, and compression opportunities in chat transcripts and AI instruction files.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| files | Files or glob patterns to audit (space-separated). Example: "CLAUDE.md conversations/*.jsonl" | yes | — |
| fail-under | Minimum grade threshold (A/B/C/D/F). The action fails with exit code 1 if any audited file scores below this threshold. | no | C |
| format | Output format: rich (default), json, or sarif. Use sarif for GitHub Code Scanning integration. | no | rich |
| version | Version of context-hygiene to install from PyPI. Default: latest. | no | "" |
Outputs
| name | description |
|---|---|
| grade | Overall grade of the worst-audited file |
| files-audited | Number of files successfully audited |