docs-review
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs documentation reviews against an internal style guide using standard file-reading and search tools like Read, Grep, and Bash. While the
allowed-toolsmetadata restricts tool access, the instructions reference GitHub MCP tools for legitimate PR review workflows.\n- [SAFE]: All external references, such as the Metabase style guide, are consistent with the vendor's context and intended purpose of the skill.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from documentation files and pull requests, creating a surface for potential indirect instructions.\n - Ingestion points: Documentation markdown files (
docs/**/*.md) and pull request diffs via GitHub MCP tools.\n - Boundary markers: The instructions do not define specific delimiters or security warnings to isolate the documentation content from the system logic.\n
- Capability inventory: The skill has access to
Bashand GitHub review tools to process data and post output to repositories.\n - Sanitization: No explicit sanitization or validation is implemented for the ingested markdown content before it is reviewed by the agent.
Audit Metadata