The Agent Skills Directory

[SAFE]: The skill utilizes local analysis scripts (scripts/change_coupling.py, scripts/file_length.py) to gather codebase metrics. These scripts use the subprocess module to call git commands safely by passing arguments as a list, which prevents shell injection. All inputs are sourced from user-provided command-line arguments.
[SAFE]: The skill's workflow, defined in SKILL.md, emphasizes safety and behavior preservation. It instructs the agent to pin behavior with tests before refactoring, keep diffs narrow, and verify the necessity of compatibility shims before removal. This reduces the risk of accidental or malicious functional changes.
[INDIRECT_PROMPT_INJECTION]: The skill processes local source code which could contain malicious instructions designed to influence the agent's behavior during refactoring. This is a standard surface for code analysis tools and is mitigated by the skill's mandatory review and testing workflow.
Ingestion points: The skill reads source files and git history in the repository to identify code smells and perform refactoring (as described in SKILL.md).
Boundary markers: The instructions emphasize pinning behavior with tests and reviewing diffs, but do not explicitly warn the agent to ignore instructions embedded within the code being analyzed.
Capability inventory: The skill utilizes scripts that execute git commands and is designed to modify files in the repository.
Sanitization: No specific sanitization or filtering of instructions found within the processed source code is mentioned.

improve-code-quality