Memory LanceDB Setup

Configure OpenClaw's memory-lancedb plugin to enable local semantic vector memory using LanceDB and an OpenAI-compatible embedding provider.

Enables semantic vector memory in OpenClaw: memories stored with memory_store are embedded and indexed locally, then recalled on-demand via memory_recall — no full-context load.

Prerequisites

• OpenClaw installed at /usr/local/lib/node_modules/openclaw
• An OpenAI-compatible embedding API key (Gemini AI Studio free key works well)

Setup Steps

1. Get a Gemini API Key (free)

Go to aistudio.google.com → Get API key → Create API key.

2. Configure the plugin

bash

openclaw config set plugins.entries.memory-lancedb.enabled true
openclaw config set plugins.entries.memory-lancedb.config.embedding.baseUrl "https://generativelanguage.googleapis.com/v1beta/openai/"
openclaw config set plugins.entries.memory-lancedb.config.embedding.model "text-embedding-004"
openclaw config set plugins.entries.memory-lancedb.config.embedding.apiKey "YOUR_API_KEY"
openclaw config set plugins.entries.memory-lancedb.config.embedding.dimensions 768

3. Install dependencies

bash

# Step 1: install main package in openclaw root
cd /usr/local/lib/node_modules/openclaw
npm install @lancedb/lancedb

# Step 2: install platform-specific native binding in plugin dir
cd /usr/local/lib/node_modules/openclaw/extensions/memory-lancedb
npm install @lancedb/lancedb-darwin-arm64   # Apple Silicon (arm64)
# npm install @lancedb/lancedb-darwin-x64   # Intel Mac
# npm install @lancedb/lancedb-linux-x64-gnu  # Linux x64

4. Patch native.js (Apple Silicon only)

LanceDB's native.js tries x64 first, hits break on failure, and never reaches arm64. Run the patch script:

bash

python3 ~/.openclaw/workspace/skills/memory-lancedb-setup/references/patch_native.py

5. Restart gateway and verify

bash

openclaw gateway restart

Then test:

text

memory_store → should return: Stored: "..."
memory_recall → should return matching entries with similarity %

Migrating from MEMORY.md

If MEMORY.md is large, migrate key facts to the vector store and shrink MEMORY.md to a 20-30 line index. Group by topic and call memory_store for each:

• Identity & permissions
• Execution rules
• Project configurations (cron IDs, doc tokens)
• Technical knowledge (API quirks, field names)
• Workflows and SOPs

Keep only "must-know-every-session" rules in MEMORY.md.

Troubleshooting

See references/troubleshooting.md for common errors and fixes.