​
Server Information

Server Name: the_swarm
Server Type: Remote MCP Server (HTTP-based)
Endpoint: https://bee.theswarm.com/mcp
Available Functions:
  • search_profiles
  • profile_fetch
  • profile_dictionary
  • network_mapper
  • get_profile_mapping
  • network_mapping_status (*require API Partner access)
  • partner_network_mapper (*require API Partner access)
Protocol: MCP over HTTP
Authorisation: The Swarm MCP Server requires API key for authentication
For desktop use: It requires proxy tool to bridge MCP protocol with HTTP

​
What It Offers

The Swarm MCP is a comprehensive professional networking and intelligence tool that provides deep insights into professional profiles and business networks. This MCP enables users to: Core Capabilities:
  • Advanced Profile Search: Search through professional profiles using powerful OpenSearch queries with customizable filters for network relevance and result limits
  • Detailed Profile Intelligence: Fetch comprehensive professional data including employment history, education, skills, company insights, and career progression patterns - ideal for sales prospecting, recruiting, and competitive intelligence
  • Network Mapping: Discover and analyze connections within your team’s professional network, revealing relationship strengths and pathways
  • Smart Filtering: Use dictionary-based searches across industries, locations, job titles, and other professional attributes with flexible text and keyword matching
  • Schema Discovery: Access the complete data structure and field mappings to optimize search strategies
Partner-Level Features (requires Partner API access):
  • External Company Network Mapping: Map and analyze any company’s professional network by domain (e.g., competitor analysis, partnership opportunities)
  • Asynchronous Processing: Handle large-scale network mapping operations with task-based processing and status monitoring
  • Cross-Company Intelligence: Identify connections and relationships between your team and target organizations

​
How It Works

Unlike standard MCP servers that run locally on your machine, The Swarm MCP Server is a remote server that:
  • Operates over HTTP rather than local processes
  • Requires network connectivity to https://bee.theswarm.com
  • Uses HTTP headers for authentication instead of local credentials
  • Runs in the cloud, eliminating the need for local installation of the actual server
  • It requires proxy tool to bridge MCP protocol with HTTP

​
Authentication

The Swarm MCP Server requires API key authentication:
  • Header: x-api-key
  • API Key Generation: Generate your API key in your team settings on webapp platform

​
How To Integrate with AI

​
Integration with Open AI

Quick Setup Connect The Swarm’s professional profile intelligence directly to OpenAI models using the Responses API (not Chat Completions). Requirements
  • OpenAI API key with Responses API access
  • Swarm API key from team settings
  • Compatible models: GPT-4o, GPT-4.1, or o-series
Basic Integration 
from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-4.1",
    tools=[{
        "type": "mcp",
        "server_label": "the_swarm",
        "server_url": "https://bee.theswarm.com/mcp",
        "headers": {"x-api-key": "YOUR_SWARM_API_KEY"}
    }],
    input="Find AI engineers in San Francisco with 5+ years experience"
)
What You Get
  • Professional Profile Search: Find profiles by skills, location, company
  • LinkedIn Data Extraction: Get detailed info from LinkedIn URLs
  • Network Intelligence: Discover connections and warm introductions
  • Real-time Access: No local setup required - works from OpenAI’s cloud
The integration is seamless - just add The Swarm as an MCP tool and your OpenAI models gain instant access to professional networking intelligence for recruiting, sales, and business development.

​
Integration with Claude Desktop

The Swarm MCP Server lets you use relationship intelligence directly inside Claude Desktop. To enable it, you’ll need to install the mcp-remote proxy and connect it to your Swarm account with an API key. Prerequisites Before you begin, make sure you have:
  • Node.js and npm installed on your system (Learn more)
  • A valid API key from your Swarm team settings
⚠️ Note: The Swarm MCP only works with Claude Desktop. It is not supported in the browser version of Claude AI.
Step 1. Install mcp-remote Open your terminal (Mac/Linux) or Command Prompt (Windows) and run: 
npm install -g mcp-remote
This installs the proxy that connects Claude Desktop with The Swarm MCP server.
Step 2. Edit Claude Desktop Configuration Find the configuration file for Claude Desktop:
  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
Open the file in a text editor and add the following configuration: 
{
  "mcpServers": {
    "the_swarm": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://bee.theswarm.com/mcp",
        "--header",
        "x-api-key: ${AUTH_TOKEN}"
      ],
      "env": {
        "AUTH_TOKEN": "YOUR_API_KEY_HERE"
      }
    }
  }
}
👉 Replace YOUR_API_KEY_HERE with your actual API key from The Swarm → Team Settings.
Step 3. Restart Claude After saving the configuration file, restart Claude Desktop.
Step 4. Verify Installation
  1. Open Claude Desktop.
  2. Check the Tools dropdown.
  3. You should see The Swarm MCP Server listed.
To test, try requesting LinkedIn profile data from within Claude.
How It Works (Simple Breakdown)
  • npx mcp-remote → downloads and runs the MCP proxy tool.
  • HTTP Bridge → translates MCP protocol into secure HTTP requests.
  • Authentication → your Swarm API key is sent in the x-api-key header.
  • Environment variable → AUTH_TOKEN safely stores your API key.
✅ That’s it! You now have The Swarm MCP Server running inside Claude Desktop.

​
Universal MCP Integration

The Swarm MCP Server works with any AI platform that supports the Model Context Protocol (MCP) standard - an open protocol for connecting AI models to external tools and data sources. Requirements Configure MCP Server Connection:
  1. Server URL: https://bee.theswarm.com/mcp
  2. Authentication: Add header x-api-key: YOUR_SWARM_API_KEY
  3. Transport: HTTP over SSE or Streamable HTTP

​
Tools Descriptions

​
search_profiles

Search profiles by OpenSearch query. Returns a list of profile IDs that match the query. Parameters
  • query object *Required _OpenSearch query to search profiles. The content of this field is placed in the “query” field of the OpenSearch request body. It MUST be a valid JSON object that matches the OpenSearch query DSL.
  • inNetworkOnly boolean - If true, only profiles that are in the user’s network will be returned. This is useful for limiting results to profiles that are more likely to be relevant to the user.
  • limit number - The maximum number of profile IDs to return. MUST be an integer.

​
profile_fetch

Get detailed information about people by their IDs or LinkedIn usernames. This includes personal information, such as name, emails and location, their current and historical employment details, education background, social media links, skills, company intelligence, and more. It delivers rich professional insights such as exact job titles, employment dates, company sizes and industries, geographic locations, and career progression patterns that are invaluable for sales prospecting, recruiting, partnership development, and competitive intelligence. You MUST provide either profiles IDs or LinkedIn usernames. You can also provide both. Parameters
  • linkedinNames array - An array of LinkedIn usernames to fetch profiles for.
  • profilesIDs array - An array of profiles IDs to fetch data for.

​
profile_dictionary

Find values of a dictionary matching the given query. Dictionaries contain a limited set of values and are useful to filter profiles. Example dictionaries: industry, location, job title, etc.. Refer to the profiles index mapping to see if the related field is mapped as “text” or “keyword”. For text fields, you can use “AND”, “OR”, and “NOT” operators to combine multiple terms. Each term is prefix-matched (an asterisk is appended to it: <term>*). For keywords, a wildcard match will be made by wrapping the term with asterisks: *<term>*. When looking for multiple values, you can use OR operator for “text” fields. For “keyword” fields you need to call this tool multiple times. Parameters
  • field string *Required - The name of the dictionary.
  • query string *Required - The query to match against the dictionary values. Must be at least 3 characters long.

​
get_profile_mapping

Get OpenSearch mapping of profiles index. Contains the full structure of the index, including field types and analyzers. This mapping is used to understand how to query the index.

​
network_mapper

Search profiles in YOUR team’s network by OpenSearch query. Returns a list of profiles (with some basic data like name and current company), with information about how they are connected to your team. Parameters
  • query object *required - OpenSearch query to search profiles. The content of this field is placed in the “query” field of the OpenSearch request body. It MUST be a valid JSON object that matches the OpenSearch query DSL.

​
partner_network_mapper

THIS TOOL CAN ONLY BE USED IF YOU HAVE PARTNER ACCESS!
Search profiles in any company network by OpenSearch query. The target team is identified by their website domain. If the team is already mapped, returns a list of profiles (with some basic data like name and current company), with information about how they are connected to your team. If the team is not yet mapped, starts an asynchronous task to map the team and returns the task ID. You can use the “network_mapping_status” tool to check the status of the task. Once the mapping is finished, calling this tool again will get the results. Parameters
  • mappingCompanyWebsite string Required - The website domain of the target company whose team you want to map.
  • query object Required - OpenSearch query to search profiles. The content of this field is placed in the “query” field of the OpenSearch request body. It MUST be a valid JSON object that matches the OpenSearch query DSL.

​
network_mapping_status

THIS TOOL CAN ONLY BE USED IF YOU HAVE PARTNER ACCESS!
Get the current status of an asynchronous network mapping task started by the “partner_network_mapper” tool. Parameters
  • taskId string *required - The ID of the mapping task to check the status for.

​
Using the Profile Fetch Function

​
profile_fetch

Purpose: Retrieves LinkedIn profile data for specified users Input Parameters:
  • linkedinNames (required): Array of LinkedIn usernames
    • Format: Username only (e.g., “johndoe” not the full URL)
    • Maximum: 1,000 usernames per request
Example Usage: 
{
  "linkedinNames": ["johndoe", "janesmithh", "techexpert2024"]
}
Username Format

​
Common Issues

  • Connection errors: Verify internet connectivity to https://bee.theswarm.com
  • Authentication failures: Check that your API key is valid and properly configured
  • MCP Remote not found: Ensure Node.js is installed and npx is available

​
Support

For technical support, API key issues, or server access, contact The Swarm support team through your platform dashboard.

​
Rate Limits

  • Maximum 1,000 profiles per request
  • Additional rate limits may apply depending on your service plan
  • API key-based rate limiting may apply
  • Contact support for specific rate limit information

​
Sample Prompts

  1. Searching for profiles data using specific URLs
Prompt example: 
Give me professional profile of https://www.linkedin.com/in/olivier-roth
The system will profile_fetch endpoint, extract the Linkedin slug and find information about the profile in The Swarm database 
{
  `linkedinNames`: [
    `olivier-roth`
  ]
}
For Best Results:
  • Use exact LinkedIn URLs (not just names)
    You can search using profile name (and the tool should look for fields like profile_info.first_name, profile_info.last_name, and profile_info.full_name) but this might not give you the accurate results and you will need to broaden the results. We encourage to search for more specific data points like Linkedin URL in this example.
2. Searching for profiles work emails Prompt example: 
Find work emails of those profiles:
[LinkedIn URL 1]
[LinkedIn URL 2]
[LinkedIn URL 3]
For Best Results:
  • Maximum 1,000 profiles per request
3. Network Connector Prompt Template Use this prompt to find the strongest connectors to your target LinkedIn profiles through The Swarm network. 
Find the strongest connectors to these target LinkedIn profiles through my team network on The Swarm:

TARGET PROFILES:
[LinkedIn URL 1]
[LinkedIn URL 2]
[LinkedIn URL 3]

MY TEAM ID: [Your Team ID from The Swarm]

Please search my team's network using the network_mapper tool and show me:
- Who in my network has the strongest connections to these people
- The connection strength and availabl connection sources 
- Create a summary with the top connectors ranked by strength

Use the successful search approach that filters by team ID first, then searches for the target profiles within that network.
For Best Results:
  1. Use exact LinkedIn URLs (not just names)
  2. Limit to 30-50 target profiles per search for focused results
  3. Include context about why these connections matter
  4. Have your team ID ready - this is crucial for the search to work
Get your Team ID:
This guide covers the setup and basic functionality of The Swarm Remote MCP Server. For advanced configuration and troubleshooting, refer to the technical documentation or contact support.