The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the retrieval and injection of past conversation data and user-modeled conclusions into the agent's prompt context, creating a surface for indirect prompt injection.
Ingestion points: Data enters the agent's context through tools like honcho_search, honcho_context, and honcho_reasoning, as well as automated background injections in hybrid and context modes.
Boundary markers: The system is documented to escape delimiter sequences and normalize whitespace to maintain prompt integrity.
Capability inventory: The skill provides capabilities for semantic search, natural language reasoning over stored data, and persistent storage of new conclusions.
Sanitization: The implementation includes stripping XML/HTML tags from user-provided content and truncating long conclusions to prevent context corruption.
[EXTERNAL_DOWNLOADS]: The skill installs the honcho-ai Python package and connects to the honcho.dev API endpoints for its memory operations. These resources are official vendor-provided components for the Hermes platform.

honcho