excalidraw

Warn

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill includes instructions to install the excalidraw-brute-export-cli package and the Playwright browser engine via npm and npx if they are not already present. This represents the installation of third-party software from public registries at runtime.- [DATA_EXFILTRATION]: The workflow involves sending diagram data to the external service https://kroki.io via curl to generate SVG exports. While this is a common utility for diagram rendering, it involves transmitting potentially sensitive content contained in the diagrams to a remote server.- [COMMAND_EXECUTION]: The skill utilizes shell access to execute curl, python, and the excalidraw-brute-export-cli tool. This broad execution capability is used for network operations and local file processing.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes user-supplied descriptions through the $ARGUMENTS variable to generate complex Excalidraw JSON structures without sanitization.
  • Ingestion points: User-provided diagram descriptions in SKILL.md.
  • Boundary markers: No delimiters or ignore instructions are used for the input data.
  • Capability inventory: The skill has access to Bash, Write, and Read tools.
  • Sanitization: No input validation or escaping mechanisms are implemented for the interpolated user content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 2, 2026, 03:09 PM