NexusMind MCP server
Commit Nov 14, 2026Provides a shared, searchable memory layer for AI coding agents to store and retrieve project context, decisions, and conventions.
Also known as smart-coder-labs/nexus-mind
- Stars
- 2
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Nov 14, 2026
- License
- MIT
What is NexusMind MCP?
NexusMind provides a shared, searchable memory layer for AI coding agents to store and retrieve project context, decisions, and conventions.
Summary generated from the project's README
About NexusMind MCP server
NexusMind provides a self-hosted, open-source memory layer designed to help AI coding agents share project context, technical decisions, and coding conventions. Instead of losing important information within individual chat sessions, teams can use this system to store knowledge that remains accessible across different tools. It functions as a centralized repository that connects to agents like Claude Code and Cursor through the Model Context Protocol.
Developers can deploy the backend and admin panel using Docker Compose, with data stored in a SQLite database by default. The system includes an admin interface to manage users, roles, and API keys while maintaining an audit trail of activity. To connect an agent, you build the MCP server and provide a specific API key to establish communication between the client and the backend REST API.
Users interact with the memory layer through three primary tools: store_memory for saving notes or findings, search_memory for retrieving context, and list_memories for browsing recent entries. This setup allows a team to maintain a consistent knowledge graph that any compatible MCP client can query.
What you can do with it
- 1Standardize coding conventionsInstruct your AI agent to save a specific naming convention so other team members can retrieve it later.
- 2Centralize technical decisionsStore project findings and architectural choices in a searchable layer to prevent repeating discovery work.
- 3Synchronize knowledge across agentsEnsure that different AI assistants working on the same project have access to the same shared context.
- 4Manage team access and securityUse the administrative panel to control which users can access the memory layer and monitor their activity.
Written from the project's README. Every use case is checked against a quote from it.
Configuration
NEXUSMIND_API_KEY
| Name | Description | |
|---|---|---|
| NEXUSMIND_API_KEY | — |
Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablesmart-coder-labs/nexus-mind is public and reachablegithub.com/smart-coder-labs/nexus-mind ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-11-13github.com/smart-coder-labs/nexus-mind ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHubWhy shared memory
AI coding assistants are useful, but project decisions and conventions often get trapped in separate chats. NexusMind is an open-source, self-hosted memory layer for AI agents: capture team knowledge once, then search and reuse it from compatible coding tools and projects.
Use it as persistent memory for Claude Code, Cursor, and MCP-compatible clients. Keep technical decisions, conventions, findings, and project context in one place, with an admin interface and an audit trail.
What you can do
- Share context across agents. Store and retrieve memories through the Model Context Protocol (MCP), so compatible assistants can work from the same team knowledge.
- Find relevant decisions. Search memories by project, tag, or content instead of repeating the same discovery in every conversation.
- See how knowledge connects. Explore related memories and projects in a knowledge graph.
- Keep teams in control. Manage users, roles, API keys, and audit events from the admin panel.
- Run it yourself. Start the backend and admin panel with Docker Compose; data is stored in SQLite by default.
Quick start
Start a local demo with Docker Compose:
make demo
Open http://localhost:3000 and sign in with the demo key printed by the command (nm_demo_acme_admin).
Heads-up: the demo setup rebuilds and reseeds
apps/backend/data/nexusmind.db. Run it in a disposable local checkout, not against a database with data you need to keep.
For a manual development setup, environment requirements, and the MCP inspector, see Running NexusMind locally.
Connect an AI coding agent
Build the MCP server and set a demo member key:
make mcp-build
export NEXUSMIND_API_KEY=nm_demo_acme_sarah
Open Claude Code from the repository root. The included MCP configuration starts NexusMind with three tools:
| Tool | Use it to |
|---|---|
store_memory |
Save a decision, convention, finding, or project note. |
search_memory |
Find relevant shared context. |
list_memories |
Browse recent memories and filter by project. |
For example, ask your coding agent: “Remember that all REST endpoints use snake_case, and tag it as a convention.” Search for that decision later from another compatible agent.
How it works
flowchart LR
Agent[Claude Code, Cursor, or MCP client] <-->|stdio| MCP[MCP server]
MCP <-->|HTTP API and API key| API[Backend REST API]
API <--> DB[(SQLite)]
Admin[Admin panel] <-->|HTTP API| API
The MCP server connects coding agents to the backend. The admin panel uses the same API to manage team access, browse knowledge, and review activity.
Technology
| Component | Technology |
|---|---|
| Backend API | Rust, Axum, SQLite |
| Admin panel | React, TypeScript, Vite |
| Agent integration | TypeScript, Model Context Protocol |
| Local deployment | Docker Compose |
Development
# Backend
make backend
# Admin panel (in another terminal)
cd apps/admin && npm install && npm run dev
# MCP server
make mcp-build
# MCP smoke test (backend must be running)
NEXUSMIND_API_KEY=nm_demo_acme_sarah make mcp-test
See docs/RUNNING.md for prerequisites, ports, demo setup, and additional workflows.
Related projects
- nexusmind-mcp — standalone MCP server package for connecting coding agents to NexusMind.
- nexusmind-claude-plugin — Claude Code plugin with the MCP server, hooks, and skills.
License
NexusMind MCP questions
What tools can I connect to NexusMind?
You can use it with Claude Code, Cursor, and any other MCP-compatible clients.
How can I store information using the MCP server?
You can use the store_memory tool to save decisions, conventions, findings, or project notes.
Where is the data stored?
By default, the system uses SQLite for data storage.
Can I manage user access and security?
Yes, the admin panel allows you to manage users, roles, API keys, and audit events.