scan
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local repository-hosted scripts using Node.js, such as
project-config-loader.cjsanddoc-stamp-guard.cjs, to manage configuration and validate document changes. - [DYNAMIC_EXECUTION]: Utility logic for handling project metadata and ensuring consistent document updates is performed by executing scripts stored within the project's own
.claude/directory. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository content to generate its output, creating an indirect injection surface.
- Ingestion points: The scan process reads project manifests, source code files, and existing documentation artifacts.
- Boundary markers: The agent is guided by specific 'Think scopes' and 'Content Rules' provided in the target manifest files.
- Capability inventory: The skill has the capability to read project files, write to the project-reference directory, and execute local validation scripts.
- Sanitization: Risks are mitigated through mandatory 'fresh-eyes' review rounds, automated path and name verification (Glob/Grep), and an explicit requirement to redact any identified secrets or credentials.
Audit Metadata