memvid/memvid

Health v2 · maintenance only

Adoption not part of the score

16449 stars · 1412 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Memvid is a single-file memory layer for AI agents that packages data, embeddings, and search indexes into one portable .mv2 file. It replaces RAG pipelines and server-based vector databases with fast hybrid (BM25 + vector) retrieval, entity state tracking, and offline-first operation.

Use cases

• give my AI agent long-term memory without running a vector database
• replace complex RAG pipelines with a single portable file
• add semantic search to my agent's knowledge base offline
• store embeddings and indexes in one file I can commit to git
• run on-prem or air-gapped retrieval for sensitive data
• query entity state and facts extracted from documents
• connect agent memory via MCP or SDK

When to choose

• you want portable, serverless memory with no database infrastructure
• you need offline or air-gapped retrieval with low latency
• you want hybrid BM25 + vector search and temporal queries in one file
• you need crash-safe, versioned memory files you can copy or sync

When to avoid

• you need multi-user concurrent writes at scale
• you require a managed distributed vector database with horizontal scaling
• your team depends on heavy existing RAG infrastructure integrations

Facets

library · maturity active

rag vector-database search-engine nlp llm-inference mcp artificial-intelligence large-language-models databases cross-platform cli python rust self-hosted single-file-memory hybrid-search bm25 embeddings offline-first portable-memory mv2 agent-memory ai-agents retrieval-augmented-generation search nodejs

Member repositories

For agents

markdown · JSON · MCP: product_card(name="memvid/memvid")