The Agent Skills Directory

[SAFE]: The skill consists of a series of Markdown style guides and templates (e.g., Decision Records, Research Papers, Status Reports). These files provide structure and best practices for the agent's output without introducing malicious logic.
[EXTERNAL_DOWNLOADS]: The documentation references official software packages including @mermaid-js/mermaid-cli and @mermaid-js/mermaid-zenuml from the npm registry. It also refers to official documentation on GitHub and mermaid.js.org. These are well-known technology providers and do not escalate the risk profile.
[COMMAND_EXECUTION]: The skill provides instructional examples for using standard command-line tools (e.g., npm, npx, git) to render diagrams and manage documentation workflows. These examples align with the allowed-tools configuration and the skill's intended purpose.
[INDIRECT_PROMPT_INJECTION]: The templates provided in the skill represent an attack surface for indirect prompt injection as they are designed to process and format external data. However, the skill incorporates mitigation strategies, such as the use of bracketed placeholders and requirements for accessible text descriptions, which improve transparency and reduce the risk of unintended behavior.

markdown-mermaid-writing