refactoring-guide
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code from external files and possesses tools to write to the filesystem and execute commands. This creates a vulnerability surface where malicious content within the analyzed code (such as comments or strings) could influence the agent's behavior.\n
- Ingestion points: External source code files analyzed using
Read,Grep, andGlobtools.\n - Boundary markers: Absent. The skill does not provide instructions to the agent to treat external code as data or use specific delimiters to isolate it.\n
- Capability inventory: Highly capable tools including
Edit,Write, andBashare allowed as per the skill's frontmatter.\n - Sanitization: Absent. No mention of validating or escaping content from external files before processing.\n- [PROMPT_INJECTION]: The
SKILL.mdfile contains a 'Self-Evolving Skill' clause instructing the agent to 'fix this file immediately' if it discovers errors or needs workarounds. This encourages the agent to rewrite its own instructions, which could be exploited through indirect injection to persistently alter the agent's logic or remove constraints.\n- [DYNAMIC_EXECUTION]: The instruction for the agent to modify its ownSKILL.mdfile constitutes a form of self-modifying code, where the agent's future logic and behavioral rules are updated dynamically at runtime.
Audit Metadata