MegaMemory MCP
A persistent-memory MCP server for coding agents: a graph of a project's concepts, architecture and decisions that survives across sessions
- 736 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Reads project code and writes a local database
- Downloads an embedding model on first run
Install
- Works in
Manual install
npm install -g megamemory
megamemory installThe interactive installer asks for your editor and sets up the server and commands.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reads project code and writes a local database
- Downloads an embedding model on first run
This is third-party code. Review the repository files before installing.
What it does
The server gives a coding agent memory that lives across sessions. The agent reads the code itself, describes concepts in its own words and writes them into a graph, then queries what it needs before a new task. The model is the indexer: there is no syntax parsing or static analysis, and it stores concepts such as features, modules, patterns and decisions rather than code symbols. All data sits in a local SQLite database in the project directory, and embeddings run in process. It installs into several editors with one command.
Who it is for. For developers who want an agent to remember a project's architecture and decisions across sessions.
Good fit when
- The agent forgets project context between sessions
- You want to store decisions and patterns, not just code
- You need semantic search over accumulated project concepts
Not a fit when
- You need syntax parsing and a precise index of code symbols
- You cannot keep a local database in the project directory
Example request
Remember this project's key architecture decisions and check them before the next taskMCP
- Transport
- stdio
- Authentication
- not required
Limitations
The model itself fills the memory, so the quality of the graph depends on how the agent describes concepts. There is no syntax parsing or static analysis, and no precise index of code symbols. It needs Node 18 or newer, and on first run it downloads an embedding model of about twenty three megabytes. Data is stored locally in SQLite in the project directory.
How to disable. Remove the MegaMemory entry from your MCP client config and uninstall the global megamemory package via npm.
FAQ
Do I need paid keys?
No. Embeddings run in process and the model downloads automatically on first run; no paid keys are needed.
Where is data stored?
In a local SQLite database inside the project directory, one file per project.
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
An MCP server that builds a code graph of functions, classes, calls and routes and answers the agent's structural queries locally
Local shared project memory for coding agents: facts, decisions and commits available through MCP, CLI and hooks in any agent
One MCP server gives agents a code map across over 300 languages, git history, shared memory and document search with OCR.
The CodeAlive MCP server: the agent searches large codebases by meaning and exactly and follows call relations without a full read
An MCP server that turns a large codebase into a semantic feature graph for precise navigation and meaning-based search
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| MegaMemory (this card) | no | not required | 736 | |
| Context Portal | no | not required | 768 | context Portal (ConPort) keeps a project's decisions |
| Nautilus Compass | no | not required | 1,258 | an MCP server of long-term memory for coding agents |
| Mnemon | no | 717 | one local binary that gives Claude Code |
README in short
The README presents MegaMemory as an MCP server that gives a coding agent a persistent knowledge graph of the project. The loop is simple: orient, work, update the memory. The model is the indexer, it stores concepts rather than code symbols, and everything lives in SQLite in the project directory. Installation is via npm install -g megamemory and an interactive megamemory install with targets for opencode, Claude Code, Antigravity and Codex. There is a web graph explorer. MIT licensed.