docs-freshness-audit
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted documentation files and source code to identify stale content, which could theoretically contain malicious instructions targeting the auditing agent.
- Ingestion points: The content of markdown files (
.md) and source code files is read during Phase 1 (grepping for paths/keywords) and Phase 2 (full content analysis by parallel agents) as defined inSKILL.md. - Boundary markers: Absent. The prompt template in Phase 2 uses simple placeholders like
<doc>and<src candidates>without explicit structural delimiters or instructions for the agent to disregard potential commands embedded within the text being audited. - Capability inventory: The skill has broad capabilities including shell command execution (
bash), interaction with GitHub APIs via theghCLI, and the ability to update or delete files in the repository during Phase 3. - Sanitization: Absent. There is no evidence of content sanitization, filtering, or validation before the document content is passed to the LLM for meaning-based auditing.
Audit Metadata