Skip to content

overview.md

Latest commit

 

History

History
54 lines (43 loc) · 3.29 KB

overview.md

File metadata and controls

54 lines (43 loc) · 3.29 KB

Overview

graphmemory is an MCP (Model Context Protocol) server that turns a project directory into a queryable semantic knowledge base. It indexes markdown documentation, TypeScript/JavaScript source code, and all project files into six interconnected graph structures, then exposes them as 70 MCP tools + REST API + web UI.

What it does

  • Parses markdown documentation into heading-based chunks with cross-file links and fenced code block extraction
  • Parses TypeScript/JavaScript source code via tree-sitter AST — extracts functions, classes, interfaces, types, enums, and their relationships
  • Indexes all project files into a file index graph with directory hierarchy, language/MIME detection
  • Stores knowledge (facts, notes, decisions) in a dedicated knowledge graph with typed relations, file attachments, and cross-graph links
  • Tracks tasks with kanban workflow, priorities, due dates, estimates, assignees, and cross-graph links
  • Manages skills (reusable recipes/procedures) with steps, triggers, usage tracking, and cross-graph links
  • Embeds every node locally using Xenova/bge-m3 by default (no external API calls); supports per-graph models with configurable pooling, normalization, dtype, and prefixes
  • Answers search queries via hybrid search (BM25 keyword + vector cosine similarity) with BFS graph expansion
  • Watches for file changes and re-indexes incrementally in real time

Key features

Feature Description
Graph types DocGraph, CodeGraph, KnowledgeGraph, FileIndexGraph, TaskGraph, SkillGraph
70 MCP tools Full CRUD + search across all graphs, cross-graph linking, attachments, epics
Multi-project One process manages multiple projects with independent graphs
Workspaces Share knowledge/tasks/skills across related projects (e.g. microservices)
REST API Express-based HTTP API for all CRUD operations
Web UI React 19 + MUI 7 dashboard with kanban board, code browsing, search
Real-time updates WebSocket push events on every mutation
Authentication Password-based login with JWT cookies, API keys for programmatic access
Access control 5-level ACL: graph > project > workspace > server > default
File mirror Notes, tasks, skills mirrored to .notes/, .tasks/, .skills/ markdown files
Reverse import Edit mirror files in IDE — changes sync back to the graph automatically
Team management .team/ directory with team members for task assignment
Embedding API Expose the server's embedding model as a REST endpoint for other services
Remote embedding Delegate embedding to a remote GPU server via HTTP
Docker Multi-platform image (amd64 + arm64) on GHCR
npm package @graphmemory/server on npm

Transports

Transport Command Use case
HTTP serve MCP clients, multiple sessions sharing one server
REST API serve Web UI, custom integrations, scripts

Requirements

  • Node.js >= 22
  • The default embedding model (Xenova/bge-m3, ~560 MB) downloads on first startup

Repository