The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on shell commands for git operations, package management, and test execution. These are used strictly within the context of verifying code refactors and repository maintenance.
[DYNAMIC_EXECUTION]: The skill performs module import checks using language interpreters (e.g., python -c, node -e). This is a targeted execution method used solely to confirm that changes have not broken module resolution.
[INDIRECT_PROMPT_INJECTION]: The skill reads repository files that may contain instructions. Evidence chain: (1) Ingestion: Files are read via git and ripgrep in SKILL.md and helper scripts. (2) Boundaries: The 6-step verification procedure (AST diffs, symbol sets) acts as a strict protocol. (3) Capabilities: The skill executes local scripts and tests. (4) Sanitization: No specific sanitization of code content is performed. This represents a low-severity risk typical for tools processing repository data.
[SAFE]: The skill follows secure software engineering practices for refactoring, including the use of isolated git worktrees, capturing baseline states for comparison, and providing detailed rollback plans. No signs of malicious intent or data exfiltration were detected.

refactor-verify