Linked API MCP
Linked API MCP server connects your LinkedIn account to AI assistants like Claude, Cursor, and VS Code. Ask them to search for leads, send messages, analyze profiles, and much more – they'll handle it through our cloud browser, safely and automatically.
## Use cases
- **Sales automation assistant**. Ask your AI to find leads, check their profiles, and draft personalized outreach. It can search for "software engineers at companies with 50-200 employees in San Francisco", analyze their backgrounds, and suggest connection messages that actually make sense.
- **Recruitment assistant**. Let your assistant search for candidates with specific skills, review their experience, and send initial outreach. It handles the time-consuming parts while you focus on actually talking to people.
- **Conversation assistant**. Your AI can read your existing LinkedIn conversations and help you respond naturally. It understands the context of your chats, suggests relevant replies, and can even send follow-up messages.
- **Market research assistant**. Need competitor analysis? Your assistant can gather data about companies, their employees, and recent activities. Get insights about industry trends without spending hours on LinkedIn.
## Get started
To start using Linked API MCP, spend 2 minutes reading these essential guides:
1. [Installation](https://linkedapi.io/mcp/installation/) – set up MCP in Claude, Cursor, VS Code, or Windsurf.
2. [Available tools](https://linkedapi.io/mcp/available-tools/) – explore all the LinkedIn tools your assistant can call.
3. [Usage examples](https://linkedapi.io/mcp/usage-examples/) – see real-world examples to get you started quickly.
## Long-running actions
Linked API actions run through a cloud browser and are queued like normal automation. Many actions take several minutes, especially searches and profile fetches with optional data.
If a tool returns `status`, `workflowId`, `operationName`, and `message`, the action is still running. Do not retry the original tool because that can queue duplicate work. Call `get_workflow_result` with the exact `workflowId` and `operationName` until the final result is returned. By default `get_workflow_result` long-polls until the workflow completes or the current MCP client's request budget elapses; pass `waitSeconds: 0` for an immediate single-shot snapshot.
## License
This project is licensed under the MIT – see the [LICENSE](https://github.com/Linked-API/linkedapi-mcp/blob/main/LICENSE) file for details.
TDQS
Scored across 59 tools
Most tools have clearly distinct targets and actions, with the nv_ prefix effectively separating Sales Navigator from standard LinkedIn operations. However, a few pairs like get_api_usage vs. admin_get_limits_usage and get_inbox vs. get_conversation could still cause initial confusion despite clear descriptions.
All tool names follow a snake_case verb_noun pattern, and the admin_ prefix is applied consistently. The main inconsistency is the mix of retrieval verbs (get, fetch, retrieve) which are used based on subtle semantic differences, slightly reducing predictability.
With 59 tools, the server vastly exceeds the 25+ threshold and clearly qualifies as an extreme mismatch per the rubric. The count is overwhelming, especially with many Sales Navigator duplicates, making it impractical for agents to select from efficiently.
The tool set provides comprehensive coverage for connection lifecycle, messaging, search, content interaction, analytics, and admin functions. Missing delete/update operations for posts or comments are minor gaps, but core workflows are well supported and no critical dead ends exist.