MGI-Mind MCP server

MGI-Mind

English | Русский | 中文

Latest release: v2.8.2 · CHANGELOG · Discussions · Issues · Contributing

Local long-term memory for AI assistants. One Rust binary, a local Qdrant vector database, local ONNX models. Speaks MCP, so Claude Code and other assistants read and write memory on their own. Also a normal CLI.

Store a fact in plain words, recall it with a query that shares none of them. Semantic search matches intent. Your assistant runs the same mind_search on its own over MCP. Regenerate with vhs docs/demo.tape.

The same exchange inside an assistant:

You:  what was the deploy server again?

Assistant (calls mind_search "deploy server"):
  -> "Deploy server is 10.0.0.5:8080, SSH as deploy@, key in vault"  (source: infra.md)

You:  right, thanks

Nothing leaves the box. Embeddings, search, reranking and the vault are all local. No cloud account, no API key, no telemetry.

Contents

• What it is
• The validity model
• Why bother
• Quick start
• Using it
• How it works
• Command reference
• Configuration
• Languages and the reranker
• Changing the embedding model
• Troubleshooting
• Security

What it is

MGI-Mind sits between you and your assistant. The assistant writes short notes ("memories") and facts as a conversation goes, and pulls the relevant ones back when they matter.

Retrieval, not just storage:

(multilingual-e5-base, 768 dims) for meaning, and a sparse term-frequency vector (TF-IDF, BM25-style) for exact words. A query runs both arms and fuses them with Reciprocal Rank Fusion. "The server box" finds "deploy host" through the dense arm; fossilize_replay finds the one note that contains that exact token through the sparse arm.

• Hybrid search. Each memory is stored as two vectors: a dense vector

bge-reranker-base, which reads the query and each passage together and is more accurate than comparing vectors. On by default and English-tuned; see Languages and the reranker for the trade-off on other languages.

• Cross-encoder reranking. Fused top candidates are re-scored by

for the whole session, models load once and stay warm in memory, so a lookup costs milliseconds instead of reloading on every call.

• One warm process. mgimind mcp is the MCP server itself: it runs

Around retrieval there's a knowledge graph for structured facts, per-agent session logs for cross-session continuity, and an encrypted terminal-only vault for secrets.

The validity model

Hybrid search is table stakes; most memory tools have some version of it. The part that is harder to find is what keeps the store from ossifying as it fills with old, contradictory, or self-reinforcing beliefs. These run mostly on their own:

Shortened. The full README is on GitHub.