The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script render-graphs.js invokes the dot (Graphviz) executable via execSync to generate SVG diagrams from markdown code blocks. The script passes data through standard input using a pipe, which is a secure method that avoids shell injection vulnerabilities. It requires the graphviz package to be pre-installed on the host system to function.
[INDIRECT_PROMPT_INJECTION]: The skill introduces a "Pressure Testing" methodology (detailed in testing-skills-with-subagents.md and persuasion-principles.md) designed to harden agent instructions against "rationalization" (the tendency for agents to skip steps under perceived pressure). It uses psychological principles such as Authority and Commitment to ensure agents adhere to strict technical standards like Test-Driven Development. This is an instructional engineering technique for improving reliability rather than a bypass of safety filters.
[INDIRECT_PROMPT_INJECTION]: The render-graphs.js utility serves as an ingestion point for untrusted data by parsing SKILL.md files for dot blocks. While the script processes this content locally and uses secure subprocess execution, it represents an automated processing surface for data within the skill's environment. The risk is minimized as the script is intended for manual execution by a developer during the skill creation process.

writing-skills