The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to clone a codebase from an unverified external source: https://github.com/yctimlin/mcp_excalidraw.
[REMOTE_CODE_EXECUTION]: The setup process involves executing npm ci and npm run build on the downloaded code, followed by running the application with node. This provides a mechanism for arbitrary code execution from a third-party source.
[COMMAND_EXECUTION]: Multiple shell commands are required for installation and operation, including git clone, npm ci, npm run build, and npm run canvas, all targeting external code.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text descriptions to generate visual elements. The absence of explicit boundary markers or input validation creates a surface where instructions embedded in the data could influence the agent's actions during the screenshot-iteration loop.
Ingestion points: SKILL.md (diagram descriptions processed at runtime).
Boundary markers: Absent; no delimiters or ignore-instructions are used for user input.
Capability inventory: batch_create_elements, get_canvas_screenshot, export_to_image, export_scene (subprocess calls and network operations).
Sanitization: Absent; input is directly used to plan and create canvas elements.

excalidraw