The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves processing .excalidraw files which are JSON-based and often contain user-defined text elements, creating a potential surface for indirect prompt injection.
Ingestion points: Instructions in SKILL.md direct the agent and its subagents to read and summarize the contents of .excalidraw files.
Boundary markers: The subagent task templates for reading or modifying diagrams do not include explicit delimiters or instructions to ignore potential commands embedded within the diagram's text labels.
Capability inventory: The skill possesses the ability to execute shell commands via a Python renderer (render_excalidraw.py), write files to the local system (PNG and JSON), and manage external browser instances via Playwright.
Sanitization: There is no explicit sanitization logic described for the text content found within the JSON elements before they are processed by the LLM or the rendering engine.
[EXTERNAL_DOWNLOADS]: The skill retrieves necessary libraries from well-known external sources to support its rendering capabilities.
The scripts/render_template.html file loads the official Excalidraw library from the esm.sh CDN at https://esm.sh/@excalidraw/excalidraw?bundle.
The SKILL.md file references an optional official MCP server at https://mcp.excalidraw.com.
[COMMAND_EXECUTION]: The skill requires and documents the execution of shell commands for setup and diagram validation.
Instructions include installing Python dependencies via pip and browser binaries via playwright install chromium.
The rendering workflow requires executing python render_excalidraw.py to convert JSON diagrams into PNG images for visual audit.

excalidraw