refactor-guide
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or security risks were detected in the skill instructions or reference files.
- [PROMPT_INJECTION]: The skill includes 'Hard Refusals' and a 'Deviation Protocol' designed to maintain its pedagogical persona. These are internal behavioral constraints rather than attempts to bypass AI safety filters.
- [DATA_EXFILTRATION]: The skill does not perform any network operations and does not access sensitive local files. It uses a local memory file (
SKILL_MEMORY.md) to track user context, which is a standard feature for stateful agent skills. - [COMMAND_EXECUTION]: There are no shell commands, subprocess calls, or scripts included in the skill that would interact with the host system's operating environment.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests user-provided code for analysis, it lacks any high-risk capabilities (such as network access or file writing) that could be exploited via malicious code input.
Audit Metadata