Memind

An open-source AI agent memory engine in Java that connects context across sessions, with an inspection UI and Apache 2.0 licensing.

Memind gives AI agents long-term memory for projects, user preferences and past tasks. Developers can use it to carry context between sessions while keeping links to the material each memory came from.

User and agent memory. Keep personal context separate from agent instructions, tool experience, reusable playbooks and resolved problems.
Connected context. Turn conversations, documents, images, audio and tool activity into searchable memories, with graphs linking related items and threads following ongoing topics.
Insight Tree. Condense related memories into summaries and broader patterns that agents can retrieve alongside individual facts and source evidence.
Layered retrieval. Combine keyword and vector search with relationships and time signals, with optional LLM-assisted deeper search and reranking.
Memory inspection. Use Memind UI to examine source data, extracted memories, relationships, insights and retrieval traces when an agent recalls the wrong context.

The Java engine is open source under Apache 2.0. Its server exposes ingestion and retrieval APIs, and its UI lets developers inspect the memory service. Applications can connect through REST, HTTP MCP, Java runtime APIs or SDKs for TypeScript, Python, Java, Go and Rust.

Official integrations for Claude Code, Codex, OpenClaw and Hermes recall context and capture agent activity across sessions. Deep retrieval uses an LLM to expand queries and check whether the returned context is sufficient.