Memorix MCP
Local shared project memory for coding agents: facts, decisions and commits available through MCP, CLI and hooks in any agent
- 930 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Reads project files and git history
- Installs integrations and runs a local memory process
Install
- Works in
Manual install
npm install -g memorix
memorix setup --agent claude --globalSwap claude for your agent: codex, cursor, copilot, gemini-cli, opencode, windsurf.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reads project files and git history
- Installs integrations and runs a local memory process
This is third-party code. Review the repository files before installing.
What it does
Memorix gives coding agents a shared project memory that survives new chats, IDE switches and handoffs. The memory sits next to the git project, SQLite is the canonical store, and search runs by keyword and by meaning. It keeps observations, fixes, gotchas and design rationale, and turns commits into engineering facts. One memory pool is reachable through MCP, hooks, CLI, SDK and a bundled terminal agent, and the setup command wires Memorix into your agent.
Who it is for. For developers tired of re-explaining the project to every new agent session.
Good fit when
- A past session figured something out that a new one does not know
- You work across several agents and IDEs on one project
- You need to keep the reasons behind architecture decisions between chats
Not a fit when
- You need cloud-shared team memory rather than local per-project memory
- The project is one-off and memory does not pay off
Example request
Save to project memory why we chose a queue over cron, and surface it next timeMCP
- Transport
- stdio
- Authentication
- not required
Limitations
Memory is local and tied to the git project; cross-device sync is optional and configured separately. The semantic index and LLM-backed memory formation are optional and need their own keys. Installation needs Node.js.
How to disable. Uninstall the package (npm uninstall -g memorix) and remove the added MCP entry from the agent config; the project memory folder can be deleted manually.
FAQ
Where is data stored?
Locally; the canonical store is SQLite next to the git project. Indexes are rebuildable, and cross-device sync is opt-in.
Is an LLM required?
No. LLM-backed memory formation and embeddings are optional; the basic path runs without an external model.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP layer of persistent collective memory for agents: structured storage, semantic search, a relation graph and a supersession lineage
A persistent-memory MCP server for coding agents: a graph of a project's concepts, architecture and decisions that survives across sessions
Shared memory for a team and its agents: it lives right in the repository and syncs through Git
Local debugging memory for developers and agents: commands, output, errors and working fixes stored in local SQLite
Git-native persistent project memory for agents based on Open Knowledge Format: a single Go binary, an embedded MCP server and BM25 search
GraphMind
graphmind
Local MCP server that builds a code graph and remembers project decisions, so the agent stops re-reading the repo every session
README in short
The README describes Memorix as local shared memory for coding agents that lives under the git project rather than inside one chat window. SQLite is the canonical store, with keyword search and an optional LanceDB semantic index. Memory kinds include observations, long-term memory, code memory, git-commit memory, design rationale and a review-gated knowledge workspace. Install is via npm, and memorix setup wires MCP, rules, hooks and skills into dozens of agents, with stdio as the default transport.