linkedin-mcp
Enables LinkedIn API integration for managing profiles, posts, feed, connections, and sending messages through MCP-compatible clients.
README
LinkedIn MCP Server
MCP server for LinkedIn API integration — profiles, posts, feed, and connections.
Works with Claude Desktop, Cursor, Windsurf, Claude Code, OpenCode, and any MCP-compatible client.
Features
| Tool | Description | Scope Required |
|---|---|---|
linkedin_get_user_info |
Get OpenID Connect user info | openid |
linkedin_get_my_profile |
Get full LinkedIn profile | openid + r_liteprofile |
linkedin_create_post |
Publish a LinkedIn post | w_member_social |
linkedin_list_posts |
List your/others' posts | r_member_social |
linkedin_delete_post |
Delete a LinkedIn post | w_member_social |
linkedin_get_feed |
Get your feed activity | r_member_social |
linkedin_get_connections |
Get your connections | Partner API |
linkedin_send_message |
Send a direct message | Partner API |
linkedin_search_people |
Search LinkedIn members | Partner API |
linkedin_oauth_login |
OAuth PKCE login flow | — |
Quick Start
# Install from npm (recommended)
npx @eggisatriadev/linkedin-mcp
# Or install locally
npm install @eggisatriadev/linkedin-mcp
# Set your access token
export LINKEDIN_ACCESS_TOKEN=AQX_your_token_here
# Run the server
npx @eggisatriadev/linkedin-mcp
For Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"linkedin": {
"command": "npx",
"args": ["-y", "@eggisatriadev/linkedin-mcp"],
"env": {
"LINKEDIN_ACCESS_TOKEN": "AQX_your_token_here"
}
}
}
}
For OpenCode
Add to your opencode.json:
{
"mcpServers": {
"linkedin": {
"command": "npx",
"args": ["-y", "@eggisatriadev/linkedin-mcp"],
"env": {
"LINKEDIN_ACCESS_TOKEN": "AQX_your_token_here"
}
}
}
}
Setup
1. Create a LinkedIn Developer App
- Go to https://www.linkedin.com/developers/apps
- Create a new app
- Add products:
- Sign In with LinkedIn using OpenID Connect (auto-approved)
- Share on LinkedIn (for posting capabilities)
- Note your Client ID and Client Secret
- Add
http://localhost:8080as an OAuth redirect URL
2. Generate Access Token
The easiest way is to use the OAuth 2.0 token generator in the LinkedIn Developer Portal:
- Go to your app → Auth tab
- In OAuth 2.0 settings, find the access token section
- Select scopes:
openid,profile,email,w_member_social - Generate and copy the token
3. Run the Server
# Install dependencies
npm install
# Set your token
export LINKEDIN_ACCESS_TOKEN=AQX_your_token_here
# Build and run
npm run build
npm start
4. Connect to Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"linkedin": {
"command": "node",
"args": ["/absolute/path/to/linkedin-mcp-server/dist/index.js"],
"env": {
"LINKEDIN_ACCESS_TOKEN": "AQX_your_token_here"
}
}
}
}
5. Connect to OpenCode
Add to your opencode.json (or use the MCP config):
{
"mcpServers": {
"linkedin": {
"command": "node",
"args": ["/absolute/path/to/linkedin-mcp-server/dist/index.js"],
"env": {
"LINKEDIN_ACCESS_TOKEN": "AQX_your_token_here"
}
}
}
}
OAuth PKCE Login (No Token Required)
Skip step 2 — let the server handle authentication interactively:
# Required: set your LinkedIn app credentials
export LINKEDIN_CLIENT_ID=your_client_id
export LINKEDIN_CLIENT_SECRET=your_client_secret
# Run the server (it will detect no token and guide you)
npm run build
npm start
# Or from npm package:
npx @eggisatriadev/linkedin-mcp
Then call the linkedin_oauth_login tool from your MCP client:
- port (optional): Callback server port (default:
8080) - open_browser (optional): Auto-open browser (default:
true) - timeout (optional): Max wait in ms (default:
120000)
The flow:
- Server starts a local HTTP server on port 8080
- Opens your browser to LinkedIn's authorization page
- You approve the request
- LinkedIn redirects to localhost — server captures the code
- Server exchanges the code for an access token via PKCE S256
- Token is saved to
~/.config/linkedin-mcp/config.json - All tools immediately work without further setup
Media Upload
LinkedIn supports image upload via a 3-step /rest/images flow:
- Initialize:
POST /rest/images?action=initializeUpload→ returnsuploadUrl+imageURN - Upload binary:
PUT {uploadUrl}with image data - Attach to post: Use the image URN as
media.idincreatePost()
Supported formats: JPEG, PNG, GIF (static) Max file size: 10 MB Recommended dimensions: 2048×2048px
Image upload happens automatically when you pass a media_url parameter
to linkedin_create_post. The MediaUploader handles retries on expired
upload URLs and validates format/size before uploading.
Development
# Watch mode (auto-reload)
npm run dev
# Build
npm run build
# Test with MCP Inspector
npm run inspector
Architecture
The server follows a layered architecture with four bounded contexts:
linkedin-mcp-server/
├── src/
│ ├── index.ts # Entry point: server init, tool registration, health check
│ ├── types.ts # Shared types and enums
│ ├── services/
│ │ └── linkedin-client.ts # LinkedIn API client (14+ methods, error classification)
│ ├── tools/
│ │ ├── profile.ts # Profile tools: getMyProfile, getUserInfo
│ │ ├── posts.ts # Posts tools: createPost, listPosts, deletePost
│ │ ├── network.ts # Network tools: getFeed, getConnections, sendMessage, searchPeople
│ │ └── auth.ts # Auth tool: oauthLogin
│ ├── auth/
│ │ ├── oauth.ts # PKCE OAuth 2.0 flow (RFC 7636)
│ │ ├── config.ts # Config file management
│ │ └── token-store.ts # Token persistence (env var → config file fallback)
│ └── media/
│ └── uploader.ts # 3-step image upload: init → binary → URN
├── tests/
│ ├── linkedin-client.test.ts
│ ├── tools/
│ ├── auth/
│ └── media/
├── docs/
│ └── pocket/arch/*/tech-design.md # Full technical design document
├── package.json
├── tsconfig.json
└── README.md
Token Resolution Chain
LINKEDIN_ACCESS_TOKENenv var (highest priority)~/.config/linkedin-mcp/config.json(persistent local token)- OAuth PKCE flow (if
LINKEDIN_CLIENT_ID+LINKEDIN_CLIENT_SECRETare set)
See tech-design.md for the full architecture, API contracts, data flow diagrams, and ADRs.
Contributing
We welcome contributions! See CONTRIBUTING.md for:
- Bug reports & feature requests (GitHub Issues)
- Development setup guide
- Coding standards & test requirements
- Pull request guidelines
This project adheres to a Code of Conduct.
Limitations
- Connections API: Requires LinkedIn Partner Program (not available on free tier)
- People Search: Not available via public API (needs Sales Navigator)
- Messaging API: Requires LinkedIn Messaging API (partner program)
- Feed: Limited by API restrictions
- Rate Limits: 100-500 requests/day per app on free tier
License
MIT — see LICENSE for details.
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。