product-marketing
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests information from the codebase (such as README, landing pages, and package.json) to generate drafts, which creates a surface for indirect prompt injection if those files contain hidden instructions.
- Ingestion points: Repository files including README, landing pages, marketing copy, and package.json as described in SKILL.md.
- Boundary markers: No specific delimiters or boundary instructions are used during the codebase scanning process.
- Capability inventory: The agent reads local files and writes the resulting context to
.agents/product-marketing.md. - Sanitization: No automated sanitization or filtering is implemented; the process relies on the user to review and correct the drafted output.
Audit Metadata