The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/generate_schematic.py employs subprocess.run to execute a secondary internal script, scripts/generate_schematic_ai.py. This modular approach uses a list of arguments rather than a raw shell string, effectively mitigating common command injection risks.
[EXTERNAL_DOWNLOADS]: The skill interacts with the OpenRouter API (openrouter.ai) to generate and review scientific diagrams. This interaction is a core functional component of the skill, and OpenRouter is recognized as a well-known service for LLM and image generation tasks.
[SAFE]: While the skill ingests untrusted scholarly data for evaluation (creating an attack surface for indirect prompt injection), the risk is minimal given the skill's capabilities are restricted to mathematical scoring and diagram generation via a trusted external service.
[SAFE]: Secret management follows best practices by requesting the OPENROUTER_API_KEY through environment variables rather than hardcoding credentials.

scholar-evaluation