The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill setup process involves cloning a repository from the author's GitHub account (github.com/wise-toddler/mcp_excalidraw).
[COMMAND_EXECUTION]: The setup script executes shell commands including npm install, npm run build, and bash launch.sh to initialize the canvas server.
[DATA_EXFILTRATION]: CLI scripts (e.g., scripts/healthcheck.cjs, scripts/export-elements.cjs) use fetch to interact with a server URL provided via command-line arguments, which could be directed to external endpoints.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) when processing untrusted diagram data.
Ingestion points: create_from_mermaid (Mermaid diagram strings), import_scene (canvas state files), and describe_scene (reading labels and elements from the canvas).
Boundary markers: None identified in the provided instructions or tool definitions.
Capability inventory: Element CRUD operations (creation, update, deletion) on the live canvas and local file system writes via export_scene and export-elements.cjs.
Sanitization: No explicit sanitization or validation of input data is present in the provided scripts.

excalidraw-skill