home / mcp / jira mcp server

Jira MCP Server

A Model Context Protocol server for Jira.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "cosmix-jira-mcp": {
      "command": "node",
      "args": [
        "/absolute/path/to/jira-mcp/build/index.js"
      ],
      "env": {
        "JIRA_TYPE": "cloud",
        "JIRA_BASE_URL": "your_jira_instance_url",
        "JIRA_API_TOKEN": "your_api_token",
        "JIRA_AUTH_TYPE": "basic",
        "JIRA_USER_EMAIL": "your_email"
      }
    }
  }
}

You run a dedicated MCP server that connects to Jira data, supports both Jira Cloud and Jira Server/Data Center, and delivers cleaned, relationship-aware data for AI context windows. This guide shows you how to use, install, and configure the Jira MCP Server so you can search issues, fetch epic children, read detailed issue data, and manage Jira content through the MCP interface.

How to use

You interact with the Jira MCP Server from your MCP client. Start the local MCP process you configured, then use the provided MCP endpoints to perform common Jira actions: search issues with JQL, retrieve epic children along with their comments and relationships, fetch detailed issue data with related issues, create or update issues, attach files, and add comments. Data is optimized for AI context windows, with rich content transformed into a clean payload that preserves essential relationships such as mentions, links, parent/child connections, and epic associations.

How to install

Prerequisites: Bun (v1.0.0 or higher) and a Jira account with API access.

1. Clone the Jira MCP project and install dependencies.

git clone [repository-url]
cd jira-mcp
bun install
bun run build

2. Configure the MCP server by adding the Jira MCP server entry to your MCP settings. Use the following configuration snippet as your starting point. Replace the absolute path to the built index and fill in your Jira credentials and preferences.

{
  "mcpServers": {
    "jira": {
      "command": "node",
      "args": ["/absolute/path/to/jira-mcp/build/index.js"],
      "env": {
        "JIRA_API_TOKEN": "your_api_token",
        "JIRA_BASE_URL": "your_jira_instance_url",
        "JIRA_USER_EMAIL": "your_email",
        "JIRA_TYPE": "cloud",
        "JIRA_AUTH_TYPE": "basic"
      }
    }
  }
}

3. Restart the MCP server to load the new Jira MCP configuration. If you are using a client like Cline, restart the MCP server from there; then relaunch your client so it can detect the new server.

Configuration and usage notes

What you configure: you can run a local, stdio-based Jira MCP server that executes the Node-based build of the MCP component. The configuration example above sets the runtime to node and points to the built entry file. The environment variables tailor how you authenticate against Jira and which Jira instance type you are using.

Authentication methods you can use depend on Jira type: - Jira Cloud: use API tokens with Basic authentication (Jira_AUTH_TYPE defaults to basic). - Jira Server/Data Center: use Basic Auth with a username/password or API token, or Bearer authentication with a Personal Access Token (PAT) if your Data Center version supports it.

Notes on limits and data handling: - Search results return up to 50 issues per request. - Epic children retrieval can return up to 100 issues per request. - The server cleans and transforms content for efficient AI context usage and tracks key relationships like mentions and epic associations.

Development and testing

For development and testing you can run the project with Bun’s test and dev commands.

bun test
bun run dev

To rebuild after changes, run the build command again.

bun run build

Supported MCP tools (endpoints and actions)

The Jira MCP Server exposes a set of tools that let you query and modify Jira data. Each tool corresponds to a common Jira operation.

Security and data handling

Security considerations include using secure authentication methods appropriate to your Jira type, protecting API tokens or PATs, and ensuring that attachments and content are handled via secure multipart requests. The MCP server also eliminates unnecessary metadata and optimizes content for AI context windows while preserving essential relationships and history.

Available tools

search_issues

Search Jira issues using JQL with a maximum of 50 results per request.

get_epic_children

Retrieve all child issues in an epic, including comments and relationship data, up to 100 issues per request.

get_issue

Fetch detailed information about a specific Jira issue, including comments and related issues.

create_issue

Create a new Jira issue with specified fields.

update_issue

Update fields of an existing Jira issue.

add_attachment

Attach a file to a Jira issue via multipart upload.

add_comment

Add a comment to a Jira issue with content converted to Atlassian Document Format.