The Agent Skills Directory

[SAFE]: The skill performs legitimate code refactoring tasks using local system utilities like grep and standard development toolchains including eslint, pytest, go test, and cargo test to identify and verify changes.
[SAFE]: Robust exclusion logic is implemented to protect sensitive system 'boundary' files, such as HTTP handlers, CLI entry points, and message consumers, from automated modifications. This ensures that necessary error handling at the system perimeter remains intact.
[SAFE]: The execution flow mandates a verification step that runs the project's own test suites to detect any regressions or previously hidden bugs immediately after refactoring is applied, reverting changes if tests fail.
[DATA_EXFILTRATION]: The skill does not perform any network operations to external domains or access sensitive credential files such as .env, AWS configurations, or SSH keys.
[PROMPT_INJECTION]: No instructions were found that attempt to override safety guidelines, bypass content filters, or ignore previous instructions.

cleanup-defensive