The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): The skill processes user search queries which are sent to the OpenRouter API. While this presents a theoretical surface for indirect prompt injection, it is the primary intended function of the skill and there are no high-impact capabilities (e.g., shell access) that would make this a significant threat. \n
Ingestion points: args.query in scripts/perplexity_search.py.\n
Boundary markers: Absent.\n
Capability inventory: Performs network requests to OpenRouter and provides optional file writing via the --output argument.\n
Sanitization: Absent.\n- [CREDENTIALS_UNSAFE] (SAFE): The skill uses environment variables or a .env file for its API key. No hardcoded credentials were found; only placeholders are used in the provided example and setup scripts.\n- [EXTERNAL_DOWNLOADS] (SAFE): The skill depends on the litellm Python package. It checks for its presence and provides installation instructions but does not perform automated installation or download external scripts.

perplexity-search