The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): File docs/advanced-techniques.md contains various shell command examples for tools like mitmproxy, semgrep, bandit, and trufflehog. These are provided as instructional snippets for manual security audits and do not execute automatically or use untrusted input.
[EXTERNAL_DOWNLOADS] (SAFE): The documentation mentions using curl, wget, and package managers (npm, pip) to gather API schemas or audit dependencies. These references are standard for a security audit toolkit and do not involve piped remote execution of untrusted scripts.
[NO_CODE] (SAFE): The skill consists entirely of Markdown files (.md). There are no executable scripts (.py, .js, .sh) provided in the skill package that would be run by the AI agent's runtime environment.
[DATA_EXPOSURE] (SAFE): While the documentation describes how to search for secrets (e.g., using grep for 'api_key' or 'secret'), it does so as an instructional guide for auditors looking at target codebases, not as a mechanism to exfiltrate the agent's or user's own credentials.

zz-code-recon