marketing-ideas
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read an external context file (.claude/product-marketing-context.md) to gather information about the product, which represents a surface where untrusted data could influence agent behavior.
- Ingestion points: Referenced in SKILL.md (instruction to read context file).
- Boundary markers: Absent; there are no delimiters or instructions to treat the context file content as potentially untrusted data.
- Capability inventory: None; the skill does not invoke any system tools, scripts, or network operations.
- Sanitization: Absent; the content of the context file is not validated or filtered before being processed by the agent.
Audit Metadata