The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the ripgrep binary through wrappers in scripts/search.mjs and scripts/quick-search.mjs. It uses Node.js child_process.spawn with shell: false, which is a security best practice that prevents shell-based command injection from malicious search patterns or arguments.
[EXTERNAL_DOWNLOADS]: The skill utilizes the @vscode/ripgrep npm package, which is a well-known service provided by a trusted organization (Microsoft/VS Code team). This package handles the automatic downloading and management of platform-specific binaries during the installation phase.
[DATA_EXFILTRATION]: While the tool supports searching hidden files (e.g., .env, .git/config) via the --hidden flag, this is an expected feature for a code discovery tool. No logic was found that would facilitate unauthorized data exfiltration or external network communication.
[PROMPT_INJECTION]: Indirect prompt injection analysis:
Ingestion points: Search results retrieved from file contents and displayed to the agent via search.mjs.
Boundary markers: ripgrep's standard output provides structural delimiters (file path, line number, and content separators).
Capability inventory: The skill is capable of executing the rg binary and performing file operations as permitted by the agent framework.
Sanitization: The use of shell: false protects the execution environment, and the documentation includes mandatory "Iron Laws" instructing agents to validate exact symbol locations before performing edits.
[SAFE]: No signs of obfuscation, hardcoded credentials, persistence mechanisms, or privilege escalation attempts were detected. The skill follows established security protocols for wrapping system CLI tools.

ripgrep