The Laimonade connector

Connect Laimonade to Claude and your coding agent can see the sprint it is working on. It reads the Ready column, pulls a story with its full acceptance criteria and the regression notes for the files it is about to touch, files the bugs it finds on the way, and hands finished work back for review — without anyone copying tickets into a prompt.

How do you connect the MCP server?

1. 01

   Add the server

   Add https://api.laimonade.com/mcp as a connector in Claude, or run /mcp in Claude Code and sign in when prompted.

2. 02

   Approve the connection

   You will be asked to sign in to Laimonade and choose which project or projects the connection is for. Approving grants the assistant access to those backlogs and nothing else.

3. 03

   One sign-in, the projects you grant

   A connection reaches only the projects you approved at sign-in, and can reach several. When it holds more than one, your assistant names its target on every call and every reply is stamped with the project that answered — so nothing lands in the wrong backlog unnoticed. The granted set can be changed later without reconnecting.

How does an agent work a sprint over MCP?

The server hands your assistant a short workflow when it connects, so in practice you ask for what you want in plain language. What follows is what happens underneath, because knowing it makes the replies easier to read.

1. 01

   Ask what is ready

   “What's ready this sprint?” returns a numbered menu of the Ready column. There are narrower versions for stories, bugs and improvements, and one that groups items under their epic.

2. 02

   Load the ones you want

   “Load 1 and 3” pulls those items in full — description, acceptance criteria, estimate, and the epic they sit under.

3. 03

   Get the context before changing anything

   Your assistant can ask for the regression notes covering the repos and paths it is about to edit, and for your project's own rules — the CLAUDE.md, AGENTS.md and Cursor rules your team already wrote.

4. 04

   Hand it back

   “Submit 1 for review” moves the item to In Review. Submitting asserts the acceptance criteria are met, which is why loading the item first matters.

The numbers always refer to the most recent list you asked for. Ask for a different list and the numbers change with it — if a reference ever looks wrong, ask for the menu again rather than guessing.

What happens when Laimonade asks you something?

Laimonade sometimes needs a person to decide something — a sprint carrying more than it can finish, milestones past their date with nothing tracked against them. It raises those on your board and in Slack, and where neither reaches you they simply wait.

So your assistant is handed anything still waiting when it starts a session, and can record your answer back. You are asked at most a few things, once per session, never in the middle of a task — and a question you ignore comes back next time rather than holding anything up. Only your answer is recorded: an assistant has no way to decide these on your behalf.

What will the connector not do?

• Close your work. There is no tool that marks an item done. An assistant can submit for review; deciding something is finished stays with a person.
• Give itself more to do. Unattended implementation is opt-in per item, set by a person. An assistant cannot add items to that queue, including its own.
• Reach a project you did not grant. A connection sees the backlog, sprints, roadmap and rules of the projects approved at sign-in, and nothing else.
• Read your conversation. The connector does not access Claude's memory, your chat history, your conversation summaries, or files you uploaded. See the privacy policy for what is stored.

What if something goes wrong?

For the model this all rests on, the full tool reference and the sprint workflow, see the documentation — in particular connect your coding agent for Cursor, Codex and headless setups.

Still stuck, or found a security issue? Email support@laimonade.com.