copy-editing
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists exclusively of Markdown documentation and reference files. No executable scripts, binaries, or automated tasks are included.
- [SAFE]: Analysis of the Seven Sweeps Framework and instructional content reveals no attempts to override agent safety protocols or exfiltrate sensitive data.
- [SAFE]: The skill interacts with local environment files (e.g., .agents/product-marketing-context.md) solely to retrieve brand guidelines. It lacks the necessary capabilities (network access, shell execution) to perform data exfiltration or malicious commands.
- [SAFE]: While the skill processes user-supplied text and local context files, representing an indirect prompt injection surface, it possesses no dangerous capabilities that could be exploited. Ingestion point: user prompt and .agents/product-marketing-context.md. Boundary markers: none. Capability inventory: none. Sanitization: none.
Audit Metadata