The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/generate_schematic.py uses subprocess.run to launch a local Python helper script. Similarly, scripts/validate_format.py uses subprocess.run to call external utilities like pdfinfo and pdffonts for document verification. These executions are performed without shell=True and are limited to local tool invocation.
[EXTERNAL_DOWNLOADS]: In scripts/generate_schematic_ai.py, the skill utilizes the requests library to make API calls to openrouter.ai. This connection is used to access Google Gemini models for the purpose of generating and critiquing scientific schematics, which is the primary functionality of the generation tool.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface within the AI diagram generation workflow.
Ingestion points: User-provided natural language descriptions for diagrams are ingested in scripts/generate_schematic.py (via the prompt argument) and passed to the AI generation script.
Boundary markers: The script uses a predefined instruction template (SCIENTIFIC_DIAGRAM_GUIDELINES) to wrap the user input before sending it to the LLM.
Capability inventory: The skill scripts have the ability to write to the file system (saving .png figures and .json logs) and perform network operations to AI service providers.
Sanitization: There is no evidence of input sanitization or filtering applied to the user-supplied prompt text before interpolation.

venue-templates