Scrumboy MCP
Scrumboy is a self-hosted Go task tracker with kanban boards, sprints and metrics that includes an MCP server for Cursor and Claude Code
- 448 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- A project token lets the agent create, change, archive and delete projects and tasks
- The admin_* tools work with an instance's users and roles
- The server takes tokens and session cookies over HTTP, which are easy to intercept without TLS in front
Install
- Works in
Manual install
docker run -d --name scrumboy -p 127.0.0.1:8080:8080 -v scrumboy-data:/data ghcr.io/markrai/scrumboy:latestStarts Scrumboy itself, with the interface at localhost:8080. The MCP endpoints use the same port.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- A project token lets the agent create, change, archive and delete projects and tasks
- The admin_* tools work with an instance's users and roles
- The server takes tokens and session cookies over HTTP, which are easy to intercept without TLS in front
This is third-party code. Review the repository files before installing.
What it does
Scrumboy is a project and task management system you run yourself: one Docker container or one executable, with data in SQLite. It offers customizable kanban boards, sprints, a calendar, workload and flow analytics across projects, a sticky-note wall, imports and backups, email notifications, webhooks and OIDC sign-in. The interface is translated into 23 languages. For agents Scrumboy opens an MCP server over HTTP: Streamable JSON-RPC at /mcp/rpc for Cursor and Claude Code, and the older /mcp interface with {tool, input} calls. Stdio is not supported. The server keeps no sessions and no SSE stream, and answers GET with 405. Tools cover projects, todos (create, search, move, archive, links), sprints, tags, members, workflow columns, priorities, a dashboard summary and metrics such as burndown, and administrators also get user operations. Access depends on the SCRUMBOY_MODE=full or anonymous mode. In full mode session cookies, static sb_… tokens and Scrumboy OAuth bound to /mcp/rpc all work. In anonymous mode the authenticated tools are effectively unavailable.
Who it is for. For teams and solo developers who want to keep a task tracker on their own server and let Cursor or Claude Code drive it through MCP.
Good fit when
- You want a tracker without SaaS: one ghcr.io/markrai/scrumboy:latest image and a data volume
- An agent should create tasks and move them across board columns from the terminal
- You need a sprint summary and a burndown chart for a report that the agent builds through metrics_getBurndown
- Sign-in should go through corporate OIDC, with agent access by token or OAuth
- The team needs project import and backups as JSON
Not a fit when
- You need MCP over stdio: Scrumboy serves it over HTTP only
- You use Jira, Linear or a Russian tracker and do not want to move tasks: separate MCP servers exist for them
- You want a read-only server: a token with project rights lets the agent create, change and delete
- The AGPL-3.0 license does not fit your project: it requires publishing source changes to the service
Example request
Create a task Update dependencies in the backend project and move it to the In progress columnMCP
- Transport
- http
- Authentication
- API key
Environment variables
SCRUMBOY_MODE- full enables sign-in and authenticated MCP tools, anonymous disables them
Limitations
Using MCP needs a running Scrumboy instance: Docker or a Go server built from source, default port 8080.
- HTTP transport only, with no stdio, no MCP sessions and no SSE.
- There are two call formats: legacy POST /mcp with {tool, input} and JSON-RPC at /mcp/rpc. Their OAuth boundaries differ, so stick to one per client.
- In anonymous mode authenticated tools are unavailable.
- The license is AGPL-3.0; Windows and macOS binaries exist, and the macOS ones are not Apple-signed.
How to disable. Remove the Scrumboy entry from the client's MCP settings and revoke the token. The service itself stops with docker stop scrumboy; data stays in the scrumboy-data volume.
FAQ
How do I connect Scrumboy to Claude Code or Cursor?
Point an HTTP MCP server at your instance's /mcp/rpc with an Authorization: Bearer header carrying an sb_… token. Cursor and Claude Code are the clients this path is meant for.
How does /mcp differ from /mcp/rpc?
/mcp is the older interface that takes JSON shaped like {tool, input}. /mcp/rpc is JSON-RPC 2.0 with tools/list and tools/call for ordinary MCP clients. Protocol versions 2025-03-26, 2025-06-18 and 2025-11-25 are supported.
How do I run Scrumboy?
Run docker run -d --name scrumboy -p 127.0.0.1:8080:8080 -v scrumboy-data:/data ghcr.io/markrai/scrumboy:latest. No .env file or TLS is needed to start, and data lives in SQLite.
Can I use Scrumboy without user accounts?
Yes, there is an anonymous mode with shared boards, but then the MCP tools that require sign-in are unavailable. An agent needs full mode.
What data can the MCP tools change?
Projects, todos, sprints, tags, members, workflow columns and priorities. The admin_* tools work on users, so expose them to an agent only when necessary.
Discussion
No questions or reviews yet.
Related
MCP serversPaca is a self-hosted task tracker where AI agents work as Scrum team members alongside people
Jira MCP in Go: an agent searches issues by JQL, runs sprints, comments and worklogs and sees an issue's branches and pull requests on Cloud and Server
Trello MCP server with 57 tools: cards, lists, checklists, attachments, comments, custom fields and workspaces, plus acceptance-criteria extraction
The official Basecamp CLI, with plugins for Claude Code and Codex, lets you run projects and todos from the terminal or an agent
yougile-mcp
@nebelov/yougile-mcp
An MCP server for the YouGile task tracker covering API v2: projects, boards, tasks, chats, users, stickers and webhooks
A self-hosted task manager for human and agent collaboration through MCP
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Scrumboy (this entry) | yes | API key | 448 | |
| Linear MCP Server | no | API key | 346 | an MCP server for cloud Linear, while you run Scrumboy on your own server |
| mcp-kaiten (Spring AI) | no | not required | 2 | an MCP server for the Russian cloud tracker Kaiten |
| Paca: Scrum task tracker with MCP for agents | yes | API key | 1,997 | Paca is a self-hosted task tracker where AI agents work as Scrum team |
README in short
The page presents Scrumboy as a self-hosted tracker: ways to run it (a Docker image, a source build, Windows and macOS executables, an Android app), a feature list, modes and roles. Optional setup follows: environment variables, an encryption key for 2FA, SMTP, notifications, OIDC, TLS and web push. A dedicated integrations section covers MCP over JSON-RPC and webhooks, and the end holds export and import rules and documentation links.