Jira Data Center MCP Server
MCP server that connects to Jira Data Center, enabling issue search, retrieval, creation, comments, workflow transitions, project listing, arbitrary JQL, and turning meeting notes into structured Jira stories, tasks, and bugs.
README
Jira Data Center MCP Server
A production-ready Model Context Protocol (MCP) server that connects an on-premises Jira Data Center instance to MCP-compatible clients such as GitHub Copilot Agent mode, Claude Desktop, or any other MCP host.
Built with TypeScript, the official @modelcontextprotocol/sdk, Axios, and Zod.
Features
- 🔐 Pluggable authentication: Personal Access Token (preferred) or Basic auth (username/password), selected automatically from environment variables via an
AuthProviderabstraction. - 🧰 12 MCP tools covering issue search, retrieval, creation, comments, workflow transitions, and project listing.
- 🧠
create_jira_story_from_requirements: turns raw workshop notes / Fit-Gap analysis text into structured Jira Stories, Tasks, and Bugs — ideal for going straight from meeting notes to a Jira backlog. - ✅ Zod-validated tool inputs, typed Jira REST responses, and normalized error handling.
- 📝 Leveled logging to
stderr(safe for the stdio MCP transport).
Project Structure
/src
server.ts # Entry point: wires config, auth, client, tools, and stdio transport
jira-client.ts # Axios-based Jira REST API v2 client + error normalization
auth.ts # AuthProvider abstraction (PAT / Basic)
config.ts # Env var loading & validation (Zod)
logger.ts # Leveled stderr logger
types.ts # Jira REST API response shapes
tools/
tool-helpers.ts # Shared CallToolResult helpers
get-current-user.ts
server-info.ts
search-issues.ts
get-issue.ts
create-issue.ts
add-comment.ts
transition-issue.ts
get-projects.ts
get-transitions.ts # bonus
get-issue-comments.ts # bonus
execute-jql.ts # bonus
create-story-from-requirements.ts# bonus: notes -> Jira backlog
requirements-parser.ts # heuristic notes parser used above
index.ts # registers all tools
Prerequisites
- Node.js >= 18
- A Jira Data Center instance reachable from this machine, with either:
- a Personal Access Token (Jira DC 8.14+,
Profile > Personal Access Tokens), or - a valid username + password
- a Personal Access Token (Jira DC 8.14+,
Setup
npm install
cp .env.example .env # then edit .env with your Jira URL + credentials
npm run build
npm start
For local iteration without a build step:
npm run dev
Environment Variables
| Variable | Required | Description |
|---|---|---|
JIRA_BASE_URL |
Yes | Base URL of your Jira Data Center instance, e.g. https://jira.company.com |
JIRA_PAT |
One of PAT/Basic | Personal Access Token (preferred auth method) |
JIRA_USERNAME |
One of PAT/Basic | Username for Basic auth (requires JIRA_PASSWORD) |
JIRA_PASSWORD |
One of PAT/Basic | Password for Basic auth (requires JIRA_USERNAME) |
JIRA_TIMEOUT_MS |
No (default 15000) |
HTTP request timeout in milliseconds |
JIRA_TLS_REJECT_UNAUTHORIZED |
No (default true) |
Set to false only for internal CAs without a valid chain |
LOG_LEVEL |
No (default info) |
debug | info | warn | error |
If JIRA_PAT is set it takes precedence; otherwise both JIRA_USERNAME and JIRA_PASSWORD must be set. The server refuses to start if neither is configured.
MCP Tools
| Tool | REST Call | Description |
|---|---|---|
get_current_user |
GET /rest/api/2/myself |
username, displayName, email, groups |
server_info |
GET /rest/api/2/serverInfo |
Jira version, deployment type, build number |
search_issues |
GET /rest/api/2/search |
Runs JQL, returns key/summary/status/assignee/reporter/created/updated |
get_issue |
GET /rest/api/2/issue/{key} |
Full issue details incl. comments, labels, assignee |
create_issue |
POST /rest/api/2/issue |
Creates an issue, returns its key |
add_comment |
POST /rest/api/2/issue/{key}/comment |
Adds a comment |
transition_issue |
POST /rest/api/2/issue/{key}/transitions |
Moves an issue through its workflow |
get_projects |
GET /rest/api/2/project |
Lists visible projects |
get_transitions (bonus) |
GET /rest/api/2/issue/{key}/transitions |
Lists valid transitions for an issue |
get_issue_comments (bonus) |
GET /rest/api/2/issue/{key}/comment |
Lists all comments on an issue |
execute_jql (bonus) |
GET /rest/api/2/search |
Arbitrary JQL with a configurable field set |
create_jira_story_from_requirements (bonus) |
POST /rest/api/2/issue (looped) |
Parses workshop notes / Fit-Gap text into Stories/Tasks/Bugs and bulk-creates them |
create_jira_story_from_requirements details
Two ways to use it:
- Automatic parsing — pass raw
notestext. The built-in heuristic parser detects:- Explicit tags: lines starting with
Story:,Task:, orBug: - User-story phrasing:
As a <role>, I want <goal> so that <benefit>→ Story - Bullet / numbered list lines → Task
- Falls back to a single Task if nothing else matches, so non-empty notes always produce at least one item.
- Explicit tags: lines starting with
- Pre-structured input — pass an
itemsarray ({ type, summary, description?, acceptanceCriteria? }) when the calling agent has already analyzed the notes itself.itemsalways takes precedence overnotes.
Set dryRun: true to preview the parsed/would-create items without touching Jira — recommended before bulk-creating from a large set of notes.
Issue creation is done per-item with Promise.allSettled, so partial failures (e.g. one bad issue type) don't block the rest; the response reports created and failed separately.
VS Code MCP Configuration
Add to your VS Code MCP configuration (e.g. .vscode/mcp.json in a workspace, or the user-level MCP settings):
{
"servers": {
"jira-datacenter": {
"type": "stdio",
"command": "node",
"args": ["${workspaceFolder}/dist/server.js"],
"env": {
"JIRA_BASE_URL": "https://jira.company.com",
"JIRA_PAT": "${input:jiraPat}"
}
}
},
"inputs": [
{
"id": "jiraPat",
"type": "promptString",
"description": "Jira Personal Access Token",
"password": true
}
]
}
Alternatively, point command at your global install (jira-mcp-server) if you npm link or npm install -g this package, or simply rely on a .env file next to dist/server.js and omit env entirely.
Example Prompts
Once connected in Copilot Agent mode:
- "Use server_info to confirm we're talking to the right Jira instance, then get_current_user to confirm my identity."
- "Search for all open bugs in project ABC assigned to me using search_issues."
- "Get the full details of ABC-123, including its comments."
- "Create a Task in project ABC titled 'Configure SSO for staging' with a short description."
- "Add a comment to ABC-123 saying the fix has been deployed to staging."
- "Show me the available transitions for ABC-123, then transition it to Done."
- "List all projects I have access to."
- "Run this JQL and show me just the priority and fixVersions fields: project = ABC AND status = 'In Progress'."
- "Here are my Fit/Gap workshop notes: [paste notes]. Preview the Jira Stories/Tasks/Bugs you'd create in project ABC with dryRun, then create them for real."
Error Handling
All Jira REST errors (4xx/5xx, network failures) are caught in jira-client.ts, mapped to a JiraApiError carrying the HTTP status code and Jira's own errorMessages/errors payload, and surfaced to the MCP client as a tool error result (isError: true) with a human-readable message — never a raw stack trace.
Security Notes
- Credentials are only ever read from environment variables — never hardcoded or logged.
- Prefer a Personal Access Token over Basic auth; PATs can be scoped and revoked independently of your account password.
- Set
JIRA_TLS_REJECT_UNAUTHORIZED=falseonly as a last resort for internal CAs; prefer installing your corporate CA certificate viaNODE_EXTRA_CA_CERTSinstead. - This server only implements the stdio MCP transport (no HTTP listener), so it is not network-exposed by itself.
- Run
npm auditperiodically — the MCP SDK's optional HTTP transport dependencies (hono,ajv/fast-uri) are not exercised by this server (stdio-only) but should still be kept current when upstream patches become available.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。