The Agent Skills Directory

[DATA_EXPOSURE]: The skill facilitates data movement by reading internal knowledge via the gbrain tool and sending it to Perplexity AI's official API (api.perplexity.ai). This behavior is consistent with the skill's documented purpose of brain-augmented research.
[CREDENTIALS_UNSAFE]: The skill requires a PERPLEXITY_API_KEY. The instructions specify that this secret should be stored in an environment file (.env), which is a standard and recommended practice for avoiding hardcoded credentials.
[INDIRECT_PROMPT_INJECTION]: The skill manages an attack surface for indirect prompt injection by processing external web content.
Ingestion points: External search results retrieved via the Perplexity API.
Boundary markers: The instructions use triple quotes and specific content headers when interpolating local context into queries to maintain separation between data and instructions.
Capability inventory: The skill utilizes gbrain put to store results in the local research/ directory.
Sanitization: Content is synthesized into a structured markdown template before being written to the filesystem.

perplexity-research