The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the node-notifier package from the public npm registry to enable its desktop alert functionality. This is a documented requirement for the background file watcher.
[PERSISTENCE]: The skill documentation provides instructions for users to add a SessionStart hook to their global ~/.claude/settings.json file. This configuration ensures the file-watcher.js process is automatically launched whenever a new Claude Code session begins.
[DYNAMIC_EXECUTION]: Helper scripts such as auto-start-watcher.js and start-watcher-interactive.js utilize Node.js spawn and execSync methods to manage and monitor the background execution of the file-watching daemon.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to read, analyze, and refactor arbitrary source code files from the user's project. This creates an attack surface where malicious instructions embedded in a processed file could potentially influence the agent's behavior.
Ingestion points: Project source code files (e.g., .ts, .tsx, .py) read during analysis or refactoring phases.
Boundary markers: None identified in the skill instructions to delimit untrusted code content.
Capability inventory: The skill directs the agent to perform file writes, version control operations (git commit), and execute project testing suites (npm run test).
Sanitization: No sanitization or safety-filtering logic is applied to the content of the files before they are processed by the agent.
[COMMAND_EXECUTION]: The skill instructions provide a workflow where the agent executes various shell commands for routine tasks, including wc for line counting, node for running alert scripts, git for backup and incremental commits, and npm for running linters and test suites.

code-refactoring