The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's documentation references well-known development tools and GitHub Actions (such as actions/checkout, coderabbitai/action, and codiumai/pr-agent) as part of a modern static analysis toolchain. These references are standard for CI/CD integration and originate from established providers.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided code. This represents a standard surface for code analysis tasks. There are no patterns indicating the skill would execute hidden instructions found within that code in a way that bypasses agent safety.
[CREDENTIALS_UNSAFE]: The implementation playbook includes educational code snippets showing examples of legacy code smells, such as hardcoded passwords, specifically to demonstrate how to refactor them into secure practices like environment variable lookups. No actual secrets are exposed.
[COMMAND_EXECUTION]: The skill describes using standard development CLI tools (e.g., ruff, mypy, eslint) for code linting and complexity analysis. These are legitimate tools for the skill's stated purpose of improving code quality.

code-refactoring-refactor-clean