The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script render-graphs.js uses child_process.execFileSync to run the host system's dot command, enabling the execution of external binary logic to process data extracted from skill files.
[PROMPT_INJECTION]: The skill instructions and the persuasion-principles.md document employ strong imperative language (e.g., 'YOU MUST', 'No exceptions', 'Never') and specific framing techniques (e.g., 'IMPORTANT: This is a real scenario') designed to override default AI behavior and enforce adherence to the TDD process.
[INDIRECT_PROMPT_INJECTION]: The render-graphs.js script provides an ingestion surface by reading and parsing SKILL.md files. It extracts content from Graphviz code blocks and passes it to an external execution tool without rigorous input sanitization, creating a potential vector for processing untrusted data.
[EXTERNAL_DOWNLOADS]: The skill's documentation within anthropic-best-practices.md includes references to external resource acquisition, specifically recommending the installation of the pdfplumber Python package via standard registry tools.

writing-skills