The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/editor_prep.py script executes the fc-match utility via subprocess.run to resolve font family names to filesystem paths. This is a legitimate use of a system utility and is implemented safely by passing a list of arguments, which prevents shell injection.
[EXTERNAL_DOWNLOADS]: The skill instructions specify installing standard, reputable Python packages through uv run, including drawsvg, svgpathtools, fonttools, cairosvg, lxml, and shapely. These are well-established libraries for vector graphics and geometry processing.
[INDIRECT_PROMPT_INJECTION]: The skill processes SVG files which may originate from external collaborators, creating a potential surface for indirect prompt injection. However, the processing is limited to structural XML transformations and does not involve executing script content or interpolating untrusted text into system-level prompts.
Ingestion points: SVG files processed by etree.parse in scripts/editor_prep.py.
Boundary markers: None explicitly defined for untrusted inputs.
Capability inventory: File system write operations and execution of the local fc-match utility.
Sanitization: Uses the lxml library for secure XML parsing and includes validation logic for font sizes and attribute units.

svg-figure