page-cro
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns were detected. The skill's functionality is entirely informational, focusing on providing expert CRO advice without the use of executable scripts, system commands, or network requests.
- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface by instructing the agent to ingest data from local project files.
- Ingestion points: .agents/product-marketing-context.md and .claude/product-marketing-context.md (SKILL.md).
- Boundary markers: Absent; instructions do not specify delimiters for the ingested content.
- Capability inventory: The skill is limited to text analysis and does not have access to tools for file modification, subprocess execution, or network access.
- Sanitization: Absent.
Audit Metadata