Honcho Memory

Honcho is an AI-native memory platform that gives Mona deep, persistent memory across sessions. Unlike simple file-based memory, Honcho uses dialectic reasoning to build a rich model of each user.

What Honcho adds

• User modeling — Honcho builds a psychological and preference model of each user
• Dialectic memory — Memory is refined through questioning and contradiction
• Multi-agent support — Multiple agents can share the same user memory
• Semantic search — Retrieve relevant memories by meaning, not just keywords

Installation

The Honcho extra is included in the [all] install. If you installed the minimal bundle:

cd ~/.monoclaw/monoclaw-runtime
uv pip install -e ".[honcho]"

Setup

1. Sign up at honcho.dev
2. Create an app and copy your API key
3. Configure MonoClaw:

monoclaw config set HONCHO_API_KEY "your-key"
monoclaw config set memory.provider honcho

Or in config.yaml:

memory:
  provider: honcho
  honcho:
    api_key: "${HONCHO_API_KEY}"
    app_id: "your-app-id"

How it works

When Honcho is enabled:

1. Every user message is analyzed for new facts, preferences, and context
2. Honcho stores these as structured memory documents
3. Before replying, Mona queries Honcho for relevant memories about the user
4. Memories are surfaced naturally in conversation — no explicit commands needed

Memory vs file-based memory

Privacy considerations

Honcho stores memory in the cloud. If you require fully local data residency, use the built-in file-based memory instead.

Troubleshooting