custom-skills-plan-analyze
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided documents (RFCs, ADRs, plans) and automatically attempts to read associated source code files mentioned within those documents. Maliciously crafted documents could contain instructions designed to influence the agent's analysis or trick it into performing unauthorized file operations.
- Ingestion points: Reads external markdown documents via the
@pathargument and reads local source code files referenced in Phase 0 and Phase 3. - Boundary markers: None identified. The skill does not implement specific delimiters or 'ignore' instructions for the content it ingests.
- Capability inventory: The agent has the ability to read arbitrary files identified in documents and can write to the file system (modifying original documents or creating new ones) in Phase 6.
- Sanitization: No sanitization or validation of the ingested document content or the resulting file-write operations is specified.
Audit Metadata