plan-stub-checker
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill operates in a 'plan-only' mode, explicitly forbidding the AI from modifying the codebase or implementing fixes until a phased plan is reviewed and approved by the user.
- [SAFE]: It utilizes a 'Preservation Contract' that prevents the deletion of features, fabrication of data/endpoints, or silent breaking changes, ensuring all findings are grounded in evidence (file:line citations).
- [COMMAND_EXECUTION]: The skill involves executing search and discovery commands (e.g.,
grep, static analysis, and optional browser-based click passes via Playwright) to inventory stubs and dead links. These are legitimate auditing actions within the skill's stated purpose. - [SAFE]: The skill uses external tools (Supabase, Sentry, Playwright) to verify integration status and backend linkages. These are well-known, trusted services used for their intended diagnostic purposes.
- [SAFE]: The instructions include a robust 'False-positive filter' phase to ensure dynamic invocations or feature-flagged code are not incorrectly flagged as dead code.
Audit Metadata