The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs subprocess.run() within generate_slide_image.py and validate_presentation.py to execute internal Python scripts and local LaTeX utilities. These operations are performed using static argument lists and avoid shell execution, preventing command injection vulnerabilities.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with the OpenRouter API (openrouter.ai) to generate images and content. This communication is restricted to the skill's stated purpose and targets a well-known service provider.\n- [PROMPT_INJECTION]: While the skill ingests user-provided text to generate slide prompts, it incorporates a structured review loop where a separate LLM call evaluates the output against scientific and design quality criteria before finalization.\n- [CREDENTIALS_UNSAFE]: The skill requires an API key for OpenRouter but follows best practices by retrieving it from environment variables or .env files, ensuring that user credentials are not hardcoded or exposed within the skill's code.

scientific-slides