The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill performs standard development tasks including identifying code changes via git and applying refactoring fixes based on quality and efficiency heuristics.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted data (code changes) and has the capability to write to the file system.
Ingestion points: Output of git diff and content of recently modified files (SKILL.md).
Boundary markers: Absent. There are no explicit instructions to treat the analyzed code as data only.
Capability inventory: Shell execution of git and file system modification for applying fixes (SKILL.md).
Sanitization: Absent. Content from files is evaluated for refactoring without explicit sanitization.

simplify