optimize-skill-md
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and processing untrusted content from user-provided SKILL.md files.
- Ingestion points: The agent reads the target SKILL.md file as part of the analysis process defined in the Step 1 section of SKILL.md.
- Boundary markers: The instructions lack explicit delimiters or markers to separate the content of the ingested file from the agent's operational instructions.
- Capability inventory: The agent utilizes a SearchReplace tool to write modified content back to the target SKILL.md file as described in Step 3 of SKILL.md.
- Sanitization: The skill does not implement content sanitization or filtering to prevent potential malicious instructions within the source markdown from affecting the agent's execution flow.
Audit Metadata