The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes external commands and Python scripts using subprocess.run to perform figure processing and export operations.\n
scripts/export.py invokes the inkscape binary for high-fidelity SVG to PDF/PNG conversion.\n
examples/two-column-figure.py executes auxiliary scripts (validate_fonts.py, export.py) using the Python interpreter to verify and export the figure.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data which could contain malicious instructions or exploit parser vulnerabilities.\n
Ingestion points: scripts/compose.py reads JSON configuration files and SVG panel files; scripts/validate_fonts.py and scripts/export.py ingest the final composed SVG figure files.\n
Boundary markers: The instructions do not define explicit boundary markers or "ignore instructions" warnings to prevent the agent from being influenced by data embedded within the processed SVG or JSON files.\n
Capability inventory: The skill has the capability to write files to the local filesystem and execute external binaries through subprocess calls.\n
Sanitization: While the skill uses the lxml.etree library for SVG parsing, it does not implement specific sanitization or validation logic to verify the safety of the XML content against injection or parser-level attacks.\n- [EXTERNAL_DOWNLOADS]: The skill utilizes the uv tool to dynamically manage and fetch Python dependencies at runtime from the official Python Package Index (PyPI).\n
Fetches packages including svgutils, lxml, cairosvg, matplotlib, and numpy.

scientific-figure