The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script render-graphs.js executes the system command dot (from the Graphviz suite) and the which command using Node.js child_process.execSync. It processes content extracted from SKILL.md and pipes it into the dot binary to generate SVG diagrams.
[PROMPT_INJECTION]: Multiple documentation files, including testing-skills-with-subagents.md and examples/CLAUDE_MD_TESTING.md, employ and teach the use of 'pressure scenarios' designed to override an agent's standard decision-making processes. These patterns use injection-like triggers such as 'IMPORTANT: This is a real scenario', 'You must choose and act', and 'No exceptions' to simulate high-stakes environments and discourage rationalization or safety-based refusals.
[EXTERNAL_DOWNLOADS]: The skill's documentation (anthropic-best-practices.md) explicitly recommends the installation and use of external libraries from public registries, including pdfplumber, pypdf, pdf2image, and docx-js to extend the agent's capabilities.

writing-skills