The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses imperative language and psychological frameworks (Authority, Scarcity, Commitment) to force agent compliance with documentation rules. Files such as persuasion-principles.md and testing-skills-with-subagents.md instruct the agent to use strong directives like "YOU MUST" and "No exceptions," as well as simulated high-pressure scenarios ("IMPORTANT: This is a real scenario") to bypass its internal rationalization processes.
[COMMAND_EXECUTION]: The render-graphs.js script utilizes child_process.execSync to run the system's dot command-line utility. It processes content extracted from markdown code blocks and pipes it directly to the utility's standard input for graph rendering.
[EXTERNAL_DOWNLOADS]: The documentation includes several links and image references to mintcdn.com and platform.claude.com. These are recognized as legitimate assets from a well-known service provider associated with the agent platform and do not contribute to verdict escalation.

writing-skills