The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard shell commands including grep, awk, sort, and test to perform static code analysis and identify patterns like long methods or duplicate code.
[COMMAND_EXECUTION]: The agent is instructed to run project-defined scripts such as npm test, npm run lint, and npm run typecheck to verify that refactoring does not break existing functionality.
[COMMAND_EXECUTION]: Integration with the claudekit CLI is used to discover and inspect documentation for other specialized agents in the environment.
[PROMPT_INJECTION]: The skill parses local source code files, which is an inherent attack surface for indirect prompt injection if those files contain malicious instructions hidden in comments or strings.
Ingestion points: SKILL.md (via grep and awk analysis of codebase files)
Boundary markers: No explicit delimiters are used to wrap the code content being analyzed
Capability inventory: Shell access for code analysis and test execution
Sanitization: No content filtering or sanitization is applied to the codebase data

refactoring-expert