honcho
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a persistent memory system that interpolates stored user data into the system prompt.
- Ingestion points: Context blocks are assembled from session summaries and user representations as described in SKILL.md.
- Boundary markers: The skill documentation specifies the use of delimiter sequences to isolate injected context.
- Capability inventory: Includes tools for writing conclusions and calling reasoning LLMs.
- Sanitization: Explicitly describes a sanitization process that strips XML/HTML tags and normalizes content to prevent injection.
- [EXTERNAL_DOWNLOADS]: Specifies the
honcho-aiPython package as a prerequisite, which is the official library for the Honcho service.
Audit Metadata