doobidoo/mcp-memory-service v10.2.0

v10.2.0 - External Embedding API Support

on GitHub

🎉 External Embedding API Support

This release adds support for using external OpenAI-compatible embedding APIs instead of local models!

✨ Key Features

• 🔌 OpenAI-Compatible API Integration: Connect to any OpenAI-compatible /v1/embeddings endpoint
  • vLLM, Ollama, Text Embeddings Inference (TEI), OpenAI, and more
• ⚙️ Simple Configuration: Set up via environment variables
• 🛡️ Graceful Fallback: Automatically falls back to local models if external API unavailable
• 🔍 Automatic Dimension Detection: Detects embedding dimensions from API responses
• ✅ Backend Validation: Ensures configuration is compatible with selected storage backend

📝 Configuration

export MCP_EXTERNAL_EMBEDDING_URL=http://localhost:8890/v1/embeddings
export MCP_EXTERNAL_EMBEDDING_MODEL=nomic-embed-text
export MCP_EXTERNAL_EMBEDDING_API_KEY=sk-xxx  # Optional

⚠️ Important

Only supported with sqlite_vec backend (not compatible with hybrid or cloudflare backends).

📚 Documentation

See docs/deployment/external-embeddings.md for complete setup guide with examples for vLLM, Ollama, TEI, and OpenAI.

🧪 Testing

• 10/10 core tests passing
• Backend validation verified

🙏 Contributors

Special thanks to @isiahw1 for this excellent contribution!

Full Changelog: https://github.com/doobidoo/mcp-memory-service/blob/v10.2.0/CHANGELOG.md#1020---2026-01-28