The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements a research feature that introduces an indirect prompt injection surface.\n
Ingestion points: scripts/generate_infographic_ai.py (via research results from the Perplexity Sonar API).\n
Boundary markers: Absent. Researched facts are directly interpolated into the generation prompt in _enhance_prompt_with_research without delimiters or defensive instructions.\n
Capability inventory: The skill is granted Bash, Read, Write, and Edit tools in SKILL.md. scripts/generate_infographic.py uses subprocess.run to trigger worker scripts, and scripts/generate_infographic_ai.py performs file writes for images and logs.\n
Sanitization: Absent. There is no filtering or escaping of the research content before it is added to the model's context.\n- [COMMAND_EXECUTION]: scripts/generate_infographic.py utilizes the subprocess module to execute internal Python scripts. This is done safely by passing arguments as a list rather than a shell string, which prevents command injection.\n- [EXTERNAL_DOWNLOADS]: The skill makes network requests to the OpenRouter and Perplexity APIs to generate infographics and gather data. These are established services required for the skill's stated purpose.

infographics