Jira MCP
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
- 97 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The token acts on your behalf: the agent creates and edits issues, moves them between statuses, writes comments and logs time
- The server keeps Jira credentials in environment variables and uses them against your company's API
- The agent downloads Jira attachments into the temporary files folder on your computer
Install
- Works in
Manual install
brew install nguyenvanduocit/tap/jira-mcpInstall through Homebrew on macOS and Linux. Pass the token and Jira address as environment variables in the MCP configuration.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The token acts on your behalf: the agent creates and edits issues, moves them between statuses, writes comments and logs time
- The server keeps Jira credentials in environment variables and uses them against your company's API
- The agent downloads Jira attachments into the temporary files folder on your computer
This is third-party code. Review the repository files before installing.
What it does
Jira MCP is a Go server that connects an agent to Jira over the REST API. It runs as a local process over stdio or, with the `-http_port` flag, as a separate HTTP server at `/mcp`. Jira Cloud needs `ATLASSIAN_HOST`, `ATLASSIAN_EMAIL` and `ATLASSIAN_TOKEN` with an API token. For Jira Server and Data Center you set `ATLASSIAN_PAT`, a personal access token, instead of the email and token pair, and the server prefers it when it is set. The agent reads an issue in full: status, assignee, description, subtasks and available transitions. It searches with JQL, creates issues and subtasks, edits their fields, moves them through the workflow by transition ID, adds links such as blocks and reads the change history. There are tools for sprints (list, active sprint, search by name), project versions, comments and worklogs. Comment and description text can be sent as Markdown, and the server converts it to ADF, the format Jira expects. The `jira_get_development_information` tool returns branches, pull requests and commits that GitHub, GitLab or Bitbucket attached to an issue. `jira_download_attachment` saves an attachment to a temporary folder and gives the agent the file path. The `ENABLED_TOOLS` variable narrows the list of exposed tools. Two ready prompts, `issue_development_tree` and `release_development_overview`, come with the server, along with a separate `jira-cli` program for the same operations from a terminal.
Who it is for. For developers, QA engineers and team leads who want an agent in Claude Desktop or Cursor to take over Jira routine: pick sprint issues, file bugs, log time, collect the pull requests linked to a release.
Good fit when
- You need a quick view of the active sprint on a board, with status and assignee for each issue
- A bug report has to become an issue with a description, subtasks and a link to the blocking issue
- A release needs the list of issues in a version with the branches and pull requests attached to them
- Time has to be logged to issues in bulk from notes in a chat
- The agent should only look at Jira: you restrict the tool list with the ENABLED_TOOLS variable
Not a fit when
- You need Confluence or Jira Service Management: the server's tools only cover Jira issues, sprints, versions and attachments
- Jira login goes only through OAuth or single sign-on without an API token: the server takes an email and token pair or a personal access token
- You want a hosted server with browser sign-in: Jira MCP runs on your machine and keeps the token in its environment variables
- For Jira and Confluence in one server, mcp-atlassian supports both products, see its card
Example request
Show the unassigned issues in the active sprint of project ABC and file a bug about the login failing in Safari with a subtask for verificationMCP
- Transport
- stdio, http
- Authentication
- key is optional
Environment variables
ATLASSIAN_HOSTrequired- Address of your Jira, for example https://your-company.atlassian.net
ATLASSIAN_EMAIL- Atlassian account email, needed for Jira Cloud together with ATLASSIAN_TOKEN
ATLASSIAN_TOKENsecret- Atlassian API token for Jira Cloud
ATLASSIAN_PATsecret- Personal access token for Jira Server and Data Center, takes precedence over the email and token pair
ENABLED_TOOLS- Comma-separated tool names; the server registers no others
Limitations
- An unofficial project with no tie to Atlassian. The license is MIT, and the latest release, 1.4.0, came out on April 18, 2026.
- The token sits in environment variables or in a `.env` file next to the server. The agent acts with the rights of the token owner in every project that account can reach.
- The tools include no deletion of issues or projects. The remaining ones can only be limited through the `ENABLED_TOOLS` list: the author gives an example of 15 read tools without the 8 that change data.
- The author describes HTTP mode as optional and meant for debugging; the main way to run the server is over stdio.
- The `jira-cli` program repeats only 12 operations from the MCP set, so it has no sprint search by name or issue links, for example.
How to disable. Remove the jira block from the MCP configuration in Claude Desktop or Cursor and revoke the API token on id.atlassian.com under API tokens. If you installed with Homebrew, run brew uninstall jira-mcp.
FAQ
How do I connect Jira MCP to Claude Desktop or Cursor?
Install the binary with brew install nguyenvanduocit/tap/jira-mcp or use the image ghcr.io/nguyenvanduocit/jira-mcp:latest. Then add a jira server to the MCP configuration and pass it ATLASSIAN_HOST, ATLASSIAN_EMAIL and ATLASSIAN_TOKEN. The author also suggests another route: paste the installation prompt into a chat with your assistant and let it walk you through.
Where do I get a token for Jira MCP?
For Jira Cloud you create an API token at id.atlassian.com/manage-profile/security/api-tokens. For Jira Server and Data Center you issue a personal access token in your profile and put it in ATLASSIAN_PAT together with the server address in ATLASSIAN_HOST.
Can I stop Jira MCP from changing issues?
Yes. The ENABLED_TOOLS variable takes a comma-separated list of tool names, and the server registers none of the others. The author gives a set of 15 read tools, such as jira_get_issue and jira_search_issue, without creating, editing, transitions, comments or worklogs.
How do I find sprint issues in Jira through an agent using JQL?
The jira_search_issue tool takes a JQL query and a list of fields, and jira_get_active_sprint returns the active sprint of a board or project. Together they give the sprint's issues with the fields you need.
How do I see the branches and pull requests attached to a Jira issue?
jira_get_development_information does that: it returns branches, pull requests and commits linked to the issue through GitHub, GitLab and Bitbucket integrations. The issue_development_tree prompt collects the same for an issue and all its subtasks.
Does Jira MCP work with Jira Server and Data Center?
Yes. Since release 1.4.0 of April 18, 2026, the server starts with just ATLASSIAN_PAT, without an email and token. It sends a Bearer header with that token on requests.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server for Jira and Confluence, cloud and self-hosted: search, create and edit issues and pages
Atlassian's official cloud MCP server for Jira, Confluence, Jira Service Management, Bitbucket, Compass and Loom, with skills
YouGile MCP
Yougile MCP Server
MCP server for YouGile: projects, boards, columns, tasks, users and task chats from the agent
A Go MCP server for reading Yandex Tracker and Yandex Wiki: issues, queues, sprints, comments, pages and dynamic tables
Trello MCP server with 57 tools: cards, lists, checklists, attachments, comments, custom fields and workspaces, plus acceptance-criteria extraction
yougile-mcp
@nebelov/yougile-mcp
An MCP server for the YouGile task tracker covering API v2: projects, boards, tasks, chats, users, stickers and webhooks
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Jira MCP (this entry) | no | key is optional | 97 | |
| MCP Atlassian | no | key is optional | 6,035 | one server for Jira and Confluence, with OAuth 2.0 sign-in and a multi-user mode |
| Atlassian Rovo MCP Server | yes | OAuth | 1,120 | Atlassian's official hosted server with OAuth 2.1, covering Confluence, Bitbucket and other products besides Jira |
| Loonflow | yes | key is optional | 2,094 | the MCP server of the Loonflow platform |
README in short
The repository holds a detailed description of every tool by group: issues, search, sprints, statuses, comments, worklogs, history, links, versions and development information. Next come installation through Homebrew, a binary and a Docker image, Cursor configurations for stdio and HTTP modes and a table of environment variables. Separate sections cover `jira-cli` with command examples and limiting tools through `ENABLED_TOOLS`. There are Go tests, a changelog, a secret-leak scan of the repository and a design document for converting Markdown to ADF.