better-writing
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The 'Reporting' section of the skill contains instructions that attempt to control the agent's final output behavior by mandating specific keywords ('Block' or 'Approve') and a specific table format. These directives override the agent's default summary and verdict protocols to fit the skill's specific audit workflow.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill is designed to analyze 'nearby copy', 'product terminology', and 'style guides' from external project files (SKILL.md).
- Boundary markers: There are no explicit delimiters or instructions provided to isolate the content being analyzed from the agent's operating instructions.
- Capability inventory: The agent is empowered to evaluate project files and generate reports that conclude with a workflow-affecting decision ('Block' or 'Approve').
- Sanitization: The skill lacks instructions for sanitizing or escaping content ingested from the project environment before processing it.
Audit Metadata