product-marketing-context
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill is designed to 'study the repo' by reading files such as README.md and marketing copy, which are external untrusted inputs.
- Ingestion points: SKILL.md specifies reading repository files (README, landing pages, package.json, etc.) and any existing docs.
- Boundary markers: The instructions do not define boundary markers to separate repository content from agent instructions.
- Capability inventory: The skill has file-write capabilities to .claude/product-marketing-context.md.
- Sanitization: No sanitization is mentioned for content extracted from the codebase. The risk is that malicious instructions in the README could influence the generation of the context document, which is later used by other tools.
Audit Metadata