simplify
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection when processing code modifications.\n- Ingestion points: User-modified code in the session context.\n- Boundary markers: None. The skill does not define specific delimiters to separate code content from instructions.\n- Capability inventory: Code refinement and rewriting instructions in SKILL.md.\n- Sanitization: None. The skill does not sanitize or validate input code for embedded instructions.\n- [NO_CODE]: The skill consists entirely of natural language instructions without accompanying scripts or executable files.
Audit Metadata