Claude Code on Jira: connect it today, and assign tickets to SHIP
Connect Claude Code to Jira today with Atlassian's MCP server. Then assign a ticket to SHIP and get a pull request that passed your CI, a review and a test.
Claude Code connects to Jira today through Atlassian's remote MCP server. Run claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp, then sign in with /mcp, and Claude Code can search, summarize, create and update Jira work items in your session. SHIP runs a separate loop for a Jira ticket: a plan, your CI, a review and a test, before the pull request reaches you.
Connect Claude Code to Jira with Atlassian's MCP server
Atlassian runs a hosted MCP server for its apps. Its setup guide gives Claude Code its own section. Add the server from your terminal:
claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcpThen open a Claude Code session and run /mcp to authenticate. The guide describes the sign-in as OAuth 2.1 by default, with an API token as an optional second way. You sign in with your own Atlassian account.
After the sign-in, Claude Code can work with Jira in plain language. The guide says you can search and summarize Jira, and create and update work items. Its example moves a work item to a new status and adds a comment. The full list of tools is in a separate Atlassian article.
This setup gives one developer's session access to Jira. The developer still decides what to build from the ticket, runs the checks and opens the pull request.
Assign a work item to the Claude agent in Jira
Jira can also start the work itself. Atlassian documents a Claude agent for Jira. You assign a work item to it, or mention it in a comment. It can "write code, push branches, and open GitHub pull requests".
The setup asks for more than the MCP route:
- Admin access in Jira Cloud, and permission to install Marketplace apps on the site.
- An Anthropic account with access to Claude Console and Claude Managed Agents.
- A GitHub personal access token with read and write access to repository contents and pull requests.
- A Summary, a Description and a Repository on each work item you assign.
Then, the guide says, the agent "will typically push changes to your GitHub repo and open a draft pull request from the service account". That draft is where your team's own review, CI and tests start.
See what a mission adds after the draft
A SHIP mission does the work that follows a draft before the pull request reaches you, and each stage leaves something you can check. The table follows the order in the SHIP introduction, as the loop runs today on Linear, Jira and GitHub Issues.
| Stage | What it adds | Where you see it |
|---|---|---|
| Planner | A plan written from the issue, the repository and the acceptance criteria, before any code exists | The issue |
| Builder | The change, made in an isolated sandbox, with your formatter, linter, typecheck and tests run before it pushes | The pull request |
| CI | Your own pipeline as a gate, with a failure sent back to the Builder along with its logs | The pull request's checks |
| Reviewer | A read of the diff against the plan, with the fixes it requires and the ones it only suggests | The issue |
| Preview deploy | The change on a live environment, deployed by your pipeline with your credentials | The preview URL |
| QA | The acceptance criteria exercised against that preview, with proof of what it checked | The issue and the mission's artifacts |
A failed gate sends the work back to the Builder with the feedback, so the mission continues. Across the 260 missions in SWE-in-a-team, the loop corrected 40 of the 46 runs it sent back without a person. A mission that fails the same way twice at one gate stops with Needs attention and leaves the pull request open for you (Retry limits).
You stay in control of the merge. SHIP marks the pull request ready once review and QA pass, and a person merges it unless you turn on auto-merge. A checkpoint can also hold the mission after the plan, so you read it before any code exists (Human in the loop). The full flow on Linear is in Claude Code on Linear.
Assign a Jira ticket to SHIP
Connect your Jira site and install the SHIP GitHub App on your repositories from SHIP's onboarding. Add an LLM credential, then create a project that points at one Jira project and one repository (hand-off options).
Write the ticket's acceptance criteria as a checklist, and set SHIP as the assignee. The plan, the review verdict and the QA proof appear on the ticket as the mission moves through the stages above. The pull request reaches you once your CI, the Reviewer and QA all pass.
A team that wants to build from a spec instead of a ticket can still skip the Planner:
ship delegate plan --title "Rate-limit search" --file docs/spec.mddelegate plan builds from your spec as you wrote it (Command reference). CI, the Reviewer and QA check the change the same way they do on an assigned ticket. The Jira page covers the full setup.
Questions
Can SHIP work on Jira issues today?
Yes. Connect Jira to SHIP and assign a ticket to it. SHIP plans the change, builds it, runs your CI, reviews it and tests it before the pull request reaches you.
How do I connect Claude Code to Jira?
Run claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp, open a Claude Code session and run /mcp to sign in. Claude Code can then search, summarize, create and update Jira work items through Atlassian's remote MCP server.
Does the Atlassian MCP server need an API token?
No. Atlassian's remote MCP server signs you in with OAuth 2.1 by default. Atlassian's setup guide lists an API token as an optional second way to authenticate.
Can Claude Code open a pull request from a Jira ticket?
Jira can assign a work item to a Claude agent that pushes a branch. Atlassian says the agent typically opens a draft pull request from a service account. The setup needs Jira admin access, an Anthropic account with Claude Managed Agents and a GitHub token.
How do I connect Jira to SHIP?
In SHIP's onboarding, connect your Jira site and complete its OAuth flow, install the SHIP GitHub App on your repositories, and add an LLM credential. Then create a project that points at one Jira project and one repository.