The Agent Skills Directory

[SAFE]: The skill provides instructions and tools for integrating Honcho memory into the agent environment. Findings include:
[EXTERNAL_DOWNLOADS]: Declares a dependency on the official honcho-ai Python package from a well-known service.
[PROMPT_INJECTION]: Explicitly describes a security feature called 'Memory-context sanitization' that strips tags, normalizes whitespace, and escapes delimiters to prevent malicious content in user-authored conclusions from affecting the system prompt.
[DATA_EXFILTRATION]: While the skill manages user data, it uses official API endpoints (app.honcho.dev) and standard local configuration files (~/.honcho/config.json). No unauthorized exfiltration patterns were detected.
[COMMAND_EXECUTION]: Provides CLI commands for setup and status (e.g., hermes honcho setup), which are standard for tool configuration and do not involve arbitrary code execution.

honcho