honcho-memory

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). SKILL.md describes using get_context to retrieve recent conversation history from Honcho and inject it into the LLM call; at runtime tools/get_context.py calls session.context(...).to_openai(...), which will include outsider-authored message text stored in the Honcho session (e.g., other peers/participants in the workspace), creating an indirect prompt-injection path.