The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the installation and use of the Mermaid library and CLI from official and well-known sources. Evidence includes: Installation of @mermaid-js/mermaid-cli from the NPM registry in mermaid-cli/README.md; Importing the Mermaid ESM module from the jsDelivr CDN in intro/getting-started.md; Fetching icon packs from the unpkg CDN in config/icons.md; Using community rendering services like mermaid.ink and the mermaid.live editor.
[COMMAND_EXECUTION]: The skill documents the execution of the mmdc (mermaid-cli) command-line tool and Docker containers to process diagrams. This is the core functionality of the skill. Examples include: Local execution: mmdc -i input.mmd -o output.svg; Docker execution using the official ghcr.io/mermaid-js/mermaid-cli image; Configuration of Puppeteer (used by the CLI) via puppeteerConfigFile.json to specify custom browser executable paths or disable the sandbox for specific environments.
[INDIRECT_PROMPT_INJECTION]: The documentation explicitly addresses the security risks of rendering diagrams from untrusted user input. It describes built-in security features designed to mitigate these risks: The securityLevel configuration which defaults to strict (encoding HTML tags); The sandbox security level which renders diagrams in an isolated to prevent JavaScript execution; Use of DOMPurify for sanitizing SVG output as noted in mermaid-api-sequence.mmd.

mermaid