crunchy-work
A local kanban board and docs tool that AI agents can drive over MCP. It enables project management without accounts or API keys.
README
Crunchy
Trello for coding agents. A kanban board and lean docs that live on your machine, which Claude Code, Cursor and the rest can drive over MCP.
No account. No API key. No database to set up. Free and MIT.
Status: pre-release. Feature-complete and tested; not yet published to npm.
npx crunchy-work
That's it — it creates ~/.crunchy, opens your browser, and offers to connect any agent
clients it finds. Or skip the UI entirely and give your agent the tools directly:
claude mcp add crunchy -- npx -y crunchy-work mcp
Already have clients installed? One command wires up all of them:
npx crunchy-work connect # add --dry-run to see what it would do
It finds Claude Code, Claude Desktop, Cursor, VS Code and Windsurf, writes each one's config in its own format, and backs up anything it touches.
Why
Your agent forgets. Crunchy doesn't.
The board and the docs are still there next session — so the plan lives somewhere your agent can read it, not in a chat log you've closed.
What it is
- Projects — each with one kanban board and its own markdown docs
- A board — columns and cards, drag and drop, live-updating
- Docs — markdown per project, for the context that isn't a task
- MCP — every capability is a tool, so an agent can do anything you can
What it isn't
Crunchy is deliberately small. No workspaces, no multiple boards per project, no list view, no calendar, no roles or permissions, no built-in AI chat. If you want those, they belong to Crunchy Team.
Commands
crunchy # start the app and open it in your browser
crunchy mcp # MCP server on stdio — no server, no port, no browser
crunchy connect # wire up installed agent clients
crunchy export [dir] # write everything out as markdown and JSON
Options: --port, --data, --no-open, --yes, --help, --version.
Your data is yours
Everything lives in one folder (~/.crunchy, or wherever --data points).
crunchy export writes it all out as readable markdown plus a complete JSON dump,
and deleting the folder removes every trace. Leaving should be easy.
Requirements
Node 22.5 or newer. Nothing else.
Development
npm install
npm run dev # web on :4420
npm run dev:server # api on :4421
npm test
npm run build
Licence
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.