Jira AI 助手 MCP 服务

Copilot Jira MCP 是一个将 Jira 与 AI 助手连接的服务端工具,支持通过自然语言在 IDE、终端或 Copilot/Claude/Cursor 中查询、创建、更新和流转 Jira 工单,适用于希望减少上下文切换、提升敏捷开发效率的团队。

  • by GiDanis
  • JavaScript
  • 本地部署
  • 5 星标
  • 更新于
在 GitHub 查看 GiDanis/copilot-jira-mcp

使用说明

🎫 Copilot Jira MCP Server

Turn your Jira into an AI-powered co-pilot.

Seamlessly interact with Jira issues, search JQL, read specs in Markdown, post comments, and transition tickets directly from your terminal or IDE using natural language.

Quick Start • Features • Supported Clients • Available Tools • Examples • Marketing & Community


⚡ 10-Second Quick Start

No manual file editing required. Run the guided installer in 1 command:

macOS / Linux

curl -fsSL https://raw.githubusercontent.com/GiDanis/copilot-jira-mcp/main/setup/install.sh | bash

Windows (PowerShell)

irm https://raw.githubusercontent.com/GiDanis/copilot-jira-mcp/main/setup/install.ps1 | iex

Or using NPX (Any OS)

npx copilot-jira-mcp setup

The setup wizard will test your credentials and automatically register the server across GitHub Copilot CLI, Claude Desktop, and Cursor!


💡 Why Copilot Jira MCP?

Without Copilot Jira MCP 😫 With Copilot Jira MCP 🚀
🔀 Constantly alt-tabbing between IDE and browser 💬 Ask your AI assistant directly in your editor
📋 Manually hunting down acceptance criteria 📝 Instant Markdown-rendered specs & subtasks
🖱️ 15 clicks to move a ticket to In Progress ⚡ "Move PROJ-42 to In Progress with comment..."
🔎 Writing manual JQL in Jira search bars 🧠 AI translates plain English into precise JQL
⏳ Context loss during code reviews 🔍 Pull requirements right alongside the diff

🔄 How It Works

sequenceDiagram
    autonumber
    actor Dev as 👨‍💻 Developer
    participant AI as 🤖 AI Assistant (Copilot / Claude / Cursor)
    participant MCP as 🎫 Copilot Jira MCP Server
    participant Jira as ☁️ Atlassian Jira Cloud

    Dev->>AI: "Summarize PROJ-123 requirements and move it to In Progress"
    AI->>MCP: Call jira_get_issue(issue_key: "PROJ-123")
    MCP->>Jira: GET /rest/api/3/issue/PROJ-123
    Jira-->>MCP: Raw Atlassian Document Format (ADF)
    MCP-->>AI: Clean Markdown spec, fields & subtasks
    AI->>MCP: Call jira_transition_issue(issue_key: "PROJ-123", transition: "In Progress")
    MCP->>Jira: POST /rest/api/3/issue/PROJ-123/transitions
    Jira-->>MCP: 204 No Content (Success)
    MCP-->>AI: Transition confirmed
    AI-->>Dev: "Here are the requirements for PROJ-123. I've updated the status to In Progress!"

✨ Key Features

  • 🔍 Smart JQL Search - Search issues via natural language or raw JQL with full pagination support
  • 📝 Native ADF ➔ Markdown Parser - Converts Atlassian Document Format (ADF) into clean Markdown (code blocks, lists, tables, quotes, headings)
  • ⚡ Workflow Transitions - Advance ticket statuses (e.g., To Do ➔ In Progress ➔ Done) with optional resolution comments
  • 💬 Comment Threads - Read discussions and append new comments to issues
  • 🆕 Create & Update Issues - Create new Tasks, Bugs, and Stories; edit summaries, descriptions, priorities, and labels
  • 🔗 Subtasks & Issue Links - Inspect parent-child hierarchies and linked blocking issues
  • 📁 Project Discovery - List all accessible Jira projects and metadata
  • 👤 My Work Overview - Instant access to issues assigned to you
  • 🔌 1-Click Multi-Client Auto-Registration - Configures Copilot CLI, Claude Desktop, and Cursor automatically
  • 🔒 Enterprise-Grade Security - Credentials strictly stay in local environment variables; zero third-party telemetry

📱 Supported Clients

Client Status Configuration File
GitHub Copilot CLI ✅ Native ~/.copilot/mcp.json
Claude Desktop ✅ Native claude_desktop_config.json
Cursor IDE ✅ Native ~/.cursor/mcp.json
Windsurf & VS Code ✅ Supported Standard MCP Stdio Configuration
Antigravity / Gemini ✅ Supported Local Stdio Transport

🛠️ Available MCP Tools

Copilot Jira MCP exposes 12 comprehensive tools spanning read, write, and workflow actions:

Tool Action Description
jira_get_issue 🔍 Read Comprehensive issue details, markdown description, subtasks, links, priority
jira_search_issues 🔍 Read Search Jira issues using JQL syntax with pagination
jira_get_my_issues 🔍 Read Quick query for tickets assigned to current user, optional status filter
jira_get_comments 🔍 Read Retrieve discussion history with author, date, and formatted text
jira_get_subtasks 🔍 Read List all subtasks and child issues linked to a parent ticket
jira_get_projects 🔍 Read List all Jira projects accessible by authenticated account
jira_get_transitions 🔍 Read List valid status transitions for an issue workflow
jira_whoami 🔍 Read Test connection and display authenticated user details
jira_create_issue ✏️ Write Create new issue (Task, Bug, Story) with summary, description, priority, labels
jira_update_issue ✏️ Write Modify existing issue summary, description, priority, or labels
jira_add_comment ✏️ Write Post a new comment to an issue (plain text or markdown supported)
jira_transition_issue ⚡ Action Move an issue through workflow states (In Progress, Done, etc.)

(Legacy aliases jira_get_ticket, jira_search_tickets, and jira_get_my_tickets are maintained for 100% backwards compatibility).


💬 Interactive Query Examples

Once installed, simply talk naturally to your AI:

# In Copilot CLI, Claude Desktop, or Cursor:

> "Show me my open tickets in project PROJ"
> "What are the acceptance criteria for PROJ-1234?"
> "Add a comment to PROJ-1234: 'PR #42 is up for review.'"
> "Move PROJ-1234 to 'In Progress'"
> "Create a high priority bug: 'Payment gateway timeout on checkout'"
> "List all blockers linked to PROJ-567"
> "Who am I logged in as in Jira?"

💻 CLI Commands

The jira-mcp CLI provides interactive management commands:

# Interactive dashboard (wizard, client registration, diagnostics)
npx copilot-jira-mcp

# Guided setup wizard
jira-mcp setup

# Register with all detected MCP client configs
jira-mcp register

# Test Jira credentials and print connection info
jira-mcp test

# Start server directly (used by AI clients)
jira-mcp start

# Show help
jira-mcp --help

⚙️ Manual Configuration

If you prefer configuring manually, set these environment variables:

export JIRA_URL="https://your-company.atlassian.net"
export JIRA_EMAIL="your-email@company.com"
export JIRA_API_TOKEN="your-atlassian-api-token"

Then add this block to your client config (~/.copilot/mcp.json or claude_desktop_config.json):

{
  "mcpServers": {
    "jira": {
      "command": "npx",
      "args": ["-y", "copilot-jira-mcp", "start"],
      "env": {
        "JIRA_URL": "${JIRA_URL}",
        "JIRA_EMAIL": "${JIRA_EMAIL}",
        "JIRA_API_TOKEN": "${JIRA_API_TOKEN}"
      }
    }
  }
}

🧪 Testing & Quality Assurance

# Run unit test suite (10/10 tests pass)
npm test

# Run ESLint check
npm run lint

🔐 Security & Privacy

  • Zero Plaintext Storage: Tokens are stored strictly in user-level environment variables or .env.
  • Zero Third-Party Relays: Direct HTTPS communication with your Jira Cloud/Server instance.
  • .gitignore Enforced: Credential files are never tracked by version control.
  • See SECURITY.md for vulnerability disclosure.

📢 Spread the Word!

If Copilot Jira MCP saves you time and context-switching:


📜 License

MIT © 2026 Giuseppe Danise — see LICENSE for details.