excalidraw

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes generating an Excalidraw .excalidraw JSON from the user’s diagram request and exporting it via Kroki/local CLI, so the agent ingests outsider-authored free text (the diagram labels/process content) directly into the .excalidraw elements.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The helper script scripts/excalidraw_lib.py fetches remote Excalidraw library JSON at runtime from https://raw.githubusercontent.com/excalidraw/excalidraw-libraries/main and then parses/merges that data into the generated .excalidraw scene, meaning external content fetched at runtime directly controls the produced diagram content.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill instructs the user to install a global npm package and to run an explicit macOS "one-time" sed patch that edits the installed CLI's source file (modifying files under the global npm root), which changes the machine's state and may require elevated privileges or unsafe file edits.