The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references several external resources for its functionality, all hosted under the author's GitHub account or Docker Hub namespace:
GitHub Action: rlespinasse/drawio-export-action@v2
Docker Images: rlespinasse/drawio-export and rlespinasse/drawio-desktop-headless
Rust Library: drawio-exporter hosted at github.com/rlespinasse/drawio-exporter.
[COMMAND_EXECUTION]: The instructions include various command-line examples for local and CI/CD execution:
docker run commands with volume mounting (-v $(pwd):/data) to process local files.
Shell commands like sed, mv, and convert (ImageMagick) are suggested in the 'Advanced Section' for pre- and post-processing of diagram files.
[INDIRECT_PROMPT_INJECTION]: The skill describes an attack surface for indirect prompt injection as it is designed to process external .drawio (XML-based) files.
Ingestion points: Diagram files provided by the user and processed via Docker or GitHub Actions (e.g., diagram.drawio).
Boundary markers: None identified in the provided command examples or instructions.
Capability inventory: File system access (via Docker volumes), process execution (docker, sed, convert), and workflow automation (GitHub Actions).
Sanitization: No specific sanitization or validation of the XML content is described within the skill's instructions.

drawio-export-tools