The Agent Skills Directory

[SAFE]: The skill is a standard implementation of a persona-based retrieval-augmented generation (RAG) agent. It defines a strict linguistic style and uses specific, scoped tools for data retrieval.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes user input to formulate tool queries and then synthesizes the results of those tool calls into a final response.
Ingestion points: User queries are used to construct parameters for the retrieve_mental_models, retrieve_core_beliefs, and retrieve_transcripts tools.
Boundary markers: The skill does not explicitly define delimiters for tool outputs in the final synthesis step.
Capability inventory: The skill is limited to three domain-specific information retrieval tools. It lacks the ability to execute arbitrary code, access the file system, or perform general network operations.
Sanitization: There is no explicit sanitization of tool outputs, but the highly specific persona constraints (Greg Isenberg style) act as a significant behavioral filter on the final output.

greg-isenberg