The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (render_flowchart.py, svg2pdf.py, validate_flowchart_style.py) and a Windows batch script (batch_svg2pdf_flows.bat) to process flowchart data and convert files. These scripts use standard libraries for JSON processing and file I/O within the workspace.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided natural language or structured text to define flowchart nodes and connections. While this is an ingestion surface, the data is processed by a specific rendering script that generates static SVG code, minimizing the risk of instruction override.
Ingestion points: User descriptions, step lists, and JSON specs provided to the agent as described in SKILL.md.
Boundary markers: The instructions define structured templates for input (e.g., Node/Link lists), which helps the agent distinguish data from instructions.
Capability inventory: The skill uses pathlib for file writing and subprocess-style calls to its own Python scripts and the inkscape CLI tool.
Sanitization: The render_flowchart.py script includes an encode_entities function to sanitize non-ASCII characters into HTML numeric entities before embedding them in SVG output, which prevents character encoding issues and basic injection in the resulting SVG.

svg-flowchart