Memory Lancedb Ollama

Name: Memory Lancedb Ollama
Rating: 3.5 (1 reviews)
Author: ryanclawjr-lab

By ryanclawjr-lab 👁 271 views ▲ 0 votes

Ollama embedding support for OpenClaw memory-lancedb plugin

GitHub

Configuration Example

{
  "embedding": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "baseUrl": "http://localhost:11434"
  },
  "dbPath": "~/.openclaw/memory/lancedb",
  "autoCapture": true,
  "autoRecall": true
}

README

# PR: Add Ollama Support to memory-lancedb Plugin

## Summary

This PR adds local Ollama embedding support to the OpenClaw memory-lancedb plugin, removing the dependency on OpenAI API for embeddings.

## Changes

### 1. config.ts
- Added `EmbeddingProvider` type: `"openai" | "ollama"`
- Added Ollama model dimensions: `nomic-embed-text` (768), `mxbai-embed-large` (1024), `bge-m3` (1024)
- Added `baseUrl` config option for Ollama server
- Updated config schema to support both providers

### 2. index.ts
- Added `OllamaEmbeddings` class that calls `http://localhost:11434/api/embeddings`
- Added `EmbeddingProvider` interface
- Modified plugin registration to choose provider based on config

## New Config Format

### For Ollama (local):
```json
{
  "embedding": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "baseUrl": "http://localhost:11434"
  },
  "dbPath": "~/.openclaw/memory/lancedb",
  "autoCapture": true,
  "autoRecall": true
}
```

### For OpenAI (cloud - original):
```json
{
  "embedding": {
    "provider": "openai",
    "model": "text-embedding-3-small",
    "apiKey": "${OPENAI_API_KEY}"
  },
  "dbPath": "~/.openclaw/memory/lancedb"
}
```

## Testing

1. Verify Ollama is running with nomic-embed-text:
   ```bash
   curl http://localhost:11434/api/tags | jq '.models[].name'
   ```

2. Test embedding:
   ```bash
   curl -s http://localhost:11434/api/embeddings \
     -d '{"model": "nomic-embed-text", "prompt": "test"}' | jq '.embedding | length'
   ```

3. After applying config, test memory_recall:
   - Query: "ERC-8004 registration"
   - Query: "x402 payment flow"

## Safety Rules

- [ ] Garrus reviews before merge
- [ ] Do NOT restart gateway manually
- [ ] OpenClaw will hot-reload the memory plugin
- [ ] Rollback: revert PR if errors appear

## Rollback

If issues appear:
```bash
# Revert the config to OpenAI
git revert <commit>
# Or manually set provider back to "openai"
```

---

**Status:** Ready for review

tools