Not just search —
Graph Memory doesn't just find text — it understands how your project is structured.
Up and running in 3 steps
Install
npm install -g @graphmemory/serverServe
graphmemory serveConnect
Claude · Cursor · WindsurfAI Prompt Builder
Generate optimized system prompts for any MCP-connected AI assistant.
What makes it different
Not another RAG pipeline. A structured, interconnected project memory.
Kanban Board with AI Integration
Full kanban workflow — your AI assistant creates, moves, and links tasks while you work. Priorities, assignees, due dates, estimates, and cross-graph linking to code and docs.
- Drag-and-drop across 6 statuses
- Cross-graph links to code symbols and docs
- File mirror — edit .tasks/ in your IDE
- Optimistic locking for team concurrency
Hybrid Search Across Everything
Not keyword-only, not vector-only — both. BM25 keyword matching + vector cosine similarity fused via Reciprocal Rank Fusion, then expanded via graph BFS traversal.
- BM25 + vector + RRF fusion
- Graph-aware BFS expansion
- One query searches all 6 graphs
- Local embeddings — no API calls
File Mirror — Edit in Your IDE
Every note, task, and skill is mirrored to markdown files in your project. Edit them in VS Code, commit to git, review in PRs. Changes sync back automatically.
- .notes/ — knowledge base as markdown
- .tasks/ — tasks with YAML frontmatter
- .skills/ — procedures your team can share
- Attachments stored alongside files
Cross-Graph Links
Link a task to the code it affects. Connect a note to the doc it references. Graph Memory creates phantom proxy nodes that bridge all six graphs into one connected web.
- Link any node to any other graph
- Automatic cleanup when targets are removed
- Workspace-wide links across projects
- Navigate from code → docs → tasks → knowledge
Six Interconnected Graphs
Your project, understood as structure — not just text.
Docs
Markdown parsed into heading chunks with cross-file links and code blocks
Code
tree-sitter AST — functions, classes, interfaces, imports, relationships
Knowledge
Persistent notes and facts with typed relations and cross-graph links
Tasks
Kanban workflow with priorities, assignees, due dates, and linking
Skills
Reusable recipes with steps, triggers, and usage tracking
Files
Every project file with metadata, language detection, directory tree
Built-in Web UI
Full-featured interface included — no extra setup needed.
58 MCP Tools
Search, create, link, and manage — all from your AI assistant.
Connect Any MCP Client
Run in your project directory:
claude mcp add --transport http \
--scope project graph-memory \
http://localhost:3000/mcp/my-projectBuilt for real workflows
See how teams use Graph Memory every day.
Project Onboarding
New developer? Search docs, explore code symbols, read team notes — all from your AI assistant.
Learn more →Code Review
Review PRs with full project context. Cross-reference code with docs and architecture decisions.
Learn more →Team Knowledge Base
Capture decisions, procedures, and patterns. Team members edit notes in their IDE via file mirror.
Learn more →Incident Response
Debug production issues fast. Search code, check knowledge base, create tasks for fixes.
Learn more →Production Ready
Everything you need to ship — out of the box.
Ready to give your AI a memory?
Get started in under a minute. Free, open source, runs locally.