A local MCP server that gives AI coding agents persistent memory across sessions, reducing repetitive re-explanation and token costs from manually pasting context. It exposes 53 tools to MCP-compatible clients and uses hybrid search to retrieve relevant project details automatically.
Project overview
It captures agent tool use automatically via hooks and manages memory lifecycle features such as versioning, contradiction detection, and TTL expiry without requiring manual user effort.
Project type
MCP · AI Agent · RAG
Use cases
Knowledge Q&A · Coding & Development
Deployment
Refer to project documentation
License
Apache-2.0
Best for
Developers and AI engineers who use MCP-compatible coding agents and want to avoid re-explaining project context across sessions.
Teams looking to manage context window costs by retrieving specific semantic results instead of pasting full documentation.
Key capabilities
Records every tool use via hooks with zero manual effort required from the user.
Strips API keys, secrets, and private tags before storing memories.
Ships a stdio MCP server exposing 53 memory tools to any MCP-compatible client.
Limitations and risks
Requires a specific pinned version of the iii-engine (v0.11.2) and cannot attach to a different version.
Getting started
Installation and server startup are rated as easy because the project provides a single npx command to install and run the server.
Alternatives and comparisons
Local-first, verbatim AI memory system with high retrieval recall and pluggable storage backends for preserving conversation context in AI coding sessions.
Adds an execution layer around coding agents that provides coordinated swarms, self-learning memory, federated communications, and security guardrails.