Jira MCP: connect Jira to Claude and ChatGPT

Ask Claude or ChatGPT what’s still open in the sprint, what the latest comments on PROJ-412 say, or which bugs were reported this week. The AI searches Jira with plain words or JQL, reads the issue and answers with links.

With Deployed OS everyone connects their own Atlassian account. The AI sees the issues that person can see, and creates an issue only when someone asks and confirms the text. Unassigned, with nobody mentioned.

2 read actions · 1 write action · Works in Claude, ChatGPT and Copilot

In practice

What your team can do with Jira in Claude and ChatGPT

Engineering

Ready for stand-up in two minutes

What did I touch yesterday, what’s blocked, what’s due? JQL does the filtering, the AI writes the summary with a link to every issue.
Product

From customer call to issue

Paste your notes from a customer call and get a clear bug or story back. You check the text, and it is created in the right project with labels.
Support

Is it a known bug?

Search summaries, descriptions and comments before you escalate. If there’s a match, you get the key, status and priority to tell the customer where it stands.
Management

Status without the status meeting

Ask what’s overdue across projects, by priority and due date, before the Monday meeting. The AI reads the latest comments too, so the summary reflects what people actually wrote.
Examples

Ask it like this

  • “What’s still open in the current sprint for the APP project, by priority?”

  • “Summarise PROJ-412 and the latest comments. Who is waiting on whom?”

  • “Find bugs reported in the last 7 days that mention checkout.”

  • “Which issues assigned to me are due this week?”

  • “Is there already an issue about the login timeout on Android?”

  • “Create a bug in APP: “CSV export fails for reports over 10,000 rows”, with these steps to reproduce.”

The connector

What the Jira MCP connector does today

  • Readsjira_search_issues

    Find issues you can see by words in summary, description and comments, or by JQL, newest change first. With nothing given it lists what changed in the last 90 days. Up to 100 issues.

  • Readsjira_get_issue

    Read one issue by key: its fields, labels, description and the latest 10 comments as text, with its link.

  • Writesjira_create_issue

    Create an issue in a project with type, summary, description and labels, as you and unassigned. Nobody is mentioned, so only the project’s own notifications go out.

Need it to do more? We extend connectors for the companies that use them.

Security

Who sees what

Reads issues, creates new ones
The connector searches issues by words or JQL and reads an issue with its fields, labels, description and latest 10 comments. Its only write is a new issue, created as you and unassigned. It never changes a status, assigns, comments or deletes.
Each person’s own permissions
Everyone connects their own Atlassian account, so the AI sees the projects and issues that person can see in Jira. A project they have no access to stays closed through the AI too.
You choose who gets it
Admins decide which groups get the Jira connector, and which of them may create issues. Everyone else never sees it.
Every call logged
Every search, read and new issue is logged: who, which connector, when. Deployed OS shows reads and writes per connector and per person, and you can export the log. Access is revoked centrally, including the day someone leaves.
Setup

Set up the Jira connector in three steps

01
An admin creates an OAuth 2.0 (3LO) app in the Atlassian developer console, adds the callback URL and the Jira scopes, and turns on sharing so everyone at the company can connect. Your company owns the app, not us.
02
The admin adds the app’s Client ID and Secret to Deployed OS with your site, such as acme.atlassian.net, and chooses which groups get Jira.
03
Each person connects their own Atlassian account once, and Jira is there in their next chat.
Compared

Jira’s own MCP server or Deployed OS?

Atlassian runs its own hosted MCP server, the Atlassian Rovo MCP Server. It is generally available for Atlassian Cloud and covers Jira, Confluence and more. It is a good fit when one person or a developer wants Jira in their own assistant or editor.

Deployed OS is built for the whole company. One connection for Jira and your other tools, each person’s own login, admins choosing which groups get what, and a log of every call. The same connection works in Claude, ChatGPT and Copilot, next to your company’s skills and guardrails.

Their documentation: Atlassian’s MCP server

Questions

Jira MCP: common questions

Yes. Atlassian hosts the Atlassian Rovo MCP Server, which covers Jira and Confluence, and one person can connect it to their assistant. Deployed OS gives the whole company one connection for Jira and your other tools, with access per group and a log of every call.

It can create a new issue (type, summary, description and labels) when you ask and confirm the text, if you have Edit access. It is created as you, unassigned, with nobody mentioned. It cannot change a status, assign, comment or delete. If you want it to do more, for example comment or move issues, we can extend it for you.

Yes. Deployed OS is your company’s own MCP, so one Jira connection works in ChatGPT, Claude and Copilot. Developers who live in Claude and product managers who prefer ChatGPT search the same Jira, each with their own Atlassian account.

Yes. Ask in plain words, or give it JQL such as project = APP AND status = "In Progress". With nothing to go on, it lists the issues changed in the last 90 days. A search returns up to 100 issues.

Both, in a way. Your company registers its own Jira app in the Atlassian developer console for your Jira Cloud site, and that app connects Jira to Deployed OS. Deployed OS is your MCP server, so Claude and ChatGPT reach Jira through it.

It depends on the size of your team. Get in touch and we will walk you through it.

More connectorsAll connectors
Let’s talk

See it on your own tools.

Tell us which systems your company runs on. We’ll show you what your team can ask, and what IT gets to control.

Poyan Karimi
You’ll hear from Poyan Karimi
Co-founder · Deployed AI