plan-builder
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external research documents, PRDs, and Compass artifacts to generate project plans. It lacks specific instructions for boundary markers or sanitization of this input. As the skill has write and edit capabilities, it is vulnerable to indirect prompt injection where malicious content in source documents could influence the resulting build plan or agent behavior.\n
- Ingestion points: SKILL.md and research-extraction.md describe reading research documents and PRDs.\n
- Boundary markers: None specified for delimited ingestion.\n
- Capability inventory: Write and Edit tools are available to the skill as per SKILL.md frontmatter.\n
- Sanitization: No sanitization or validation of the input content is mentioned.
Audit Metadata