| name | honcho-caroline |
| description | Honcho-powered persistent memory and reasoning layer for Caroline AI. Use for: ingesting conversation messages into Honcho, querying Honcho's synthesised model of Wade, syncing Honcho's reasoning into caroline_neuro_memory.json, and verifying Honcho connection health. This Node follows the Centauri Interlock Standard. |
Honcho-Caroline Memory Node
This skill wires Honcho — Plastic Labs' AI-native memory and reasoning platform — directly into Caroline's memory system as a Centauri Interlock Node. Instead of storing flat facts, Honcho builds and maintains a continuously evolving psychological model of Wade across every session.
Architecture
Caroline AI (RunPod)
│
▼
honcho_caroline_node.py ◄──── caroline_neuro_memory.json
│ ▲
▼ │
Honcho API action_sync()
(Workspace: centauri-os)
├── Peer: wade
└── Peer: caroline
Core Actions
| Action | What It Does |
|---|
status | Verify Honcho connection and workspace health |
ingest | Write a conversation message (user or assistant) into Honcho |
query | Ask Honcho's reasoning engine what it knows about Wade |
sync | Pull Honcho's latest Wade-model into caroline_neuro_memory.json |
Usage
Check Connection
python3 /home/ubuntu/skills/honcho-caroline/scripts/honcho_caroline_node.py --action status
Ingest a Message (Wade speaking)
python3 /home/ubuntu/skills/honcho-caroline/scripts/honcho_caroline_node.py \
--action ingest --role user --content "I'm focused on finishing the Centauri OS today"
Ingest a Message (Caroline responding)
python3 /home/ubuntu/skills/honcho-caroline/scripts/honcho_caroline_node.py \
--action ingest --role assistant --content "Got it. I'll keep context tight on Centauri today."
Query Honcho About Wade
python3 /home/ubuntu/skills/honcho-caroline/scripts/honcho_caroline_node.py \
--action query --question "What are Wade's top priorities right now?"
Sync Honcho → caroline_neuro_memory.json
python3 /home/ubuntu/skills/honcho-caroline/scripts/honcho_caroline_node.py --action sync
Environment Variables
| Variable | Default | Description |
|---|
HONCHO_API_KEY | (set in script) | Your Honcho API key |
HONCHO_WORKSPACE | centauri-os | Honcho workspace ID |
CAROLINE_STATE_FILE | ~/wade-global-state/caroline_neuro_memory.json | Path to Centauri state file |
Integration with Centauri Stack
This Node follows the Centauri Interlock Standard:
- State Check — reads
caroline_neuro_memory.json before executing
- Modular Logic — all Honcho operations are self-contained
- Closed-Loop Broadcast — all outputs are JSON-formatted and routed to
Command_Router
Recommended Session Workflow
python3 honcho_caroline_node.py --action sync
python3 honcho_caroline_node.py --action ingest --role user --content "..."
python3 honcho_caroline_node.py --action ingest --role assistant --content "..."
python3 honcho_caroline_node.py --action query --question "What does Wade need right now?"
Honcho Workspace
- Workspace:
centauri-os
- Peers:
wade (user), caroline (assistant)
- Default Session:
caroline-memory-main
- Dashboard: app.honcho.dev