home / mcp / vector graphql mcp server

Vector GraphQL MCP Server

Queries Vector data via MCP to fetch leaderboard, profiles, trending tokens, and token broadcasts.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "jusscubs-mcp-server-test": {
      "command": "uv",
      "args": [
        "run",
        "mcp",
        "dev",
        "vector_server.py"
      ],
      "env": {
        "UV_CONFIG": "default"
      }
    }
  }
}

You can query the Vector GraphQL MCP Server to fetch leaderboard data, trader profiles, trending Solana tokens, and token broadcasts. This server exposes a simple set of tools that you can call from any MCP client to retrieve Vector’s data efficiently with MCP’s standardized interface.

How to use

To use the Vector GraphQL MCP Server, connect with your MCP client and call the four available tools: fetch_leaderboard, fetch_profile, fetch_token_data, and fetch_token_broadcasts. Use the tools to retrieve the Vector leaderboard, trader profiles, trending Solana tokens, and broadcasts for a specific token. Each tool returns structured data you can display in dashboards, integrations, or aggregations. In practice, you will invoke a tool by name and provide the required parameters through your MCP client, then use the returned data to drive your analytics, comparisons, or visualization components.

How to install

# Clone the project
git clone https://github.com/JussCubs/mcp-server-test.git
cd mcp-server-test

# Install dependencies with uv
uv add "mcp[cli]" httpx
```} ,{

If you don’t have uv installed, install it following the instructions at https://github.com/astral-sh/uv. Alternatively, you can install dependencies with Python’s package manager.

How to install (continued)

# Using pip to install requirements
pip install -r requirements.txt

Run the server in development or install mode

You can run the MCP server in development mode or install it for use on your machine. The commands shown below are the two explicit ways you can start the server.

# Development mode with the MCP Inspector
uv run mcp dev vector_server.py

# If you prefer using pip-installed tooling
mcp dev vector_server.py

Install for Claude Desktop

# Install in Claude Desktop
uv run mcp install vector_server.py

# Or if you’re using pip
mcp install vector_server.py

Troubleshooting

If you encounter a warning such as The package 'mcp==X.X.X' does not have an extra named 'cli', install the CLI tools separately.

pip install mcp httpx "click>=8.0" "rich>=10.0" "typer>=0.9.0"

Tool documentation

The server provides four tools to access Vector data.

- fetch_leaderboard: Fetches Vector leaderboard data. Parameters include leaderboard_type (default: PNL_WIN).

- fetch_profile: Fetches detailed profile data for a Vector trader. Required parameter: username.

- fetch_token_data: Fetches trending Solana tokens. This tool has no parameters and returns tokens with 5-minute broadcast activity.

- fetch_token_broadcasts: Fetches broadcasts for a specific token. Required parameter: token_id.

Available tools

fetch_leaderboard

Fetches Vector leaderboard data with a specified type (default: PNL_WIN).

fetch_profile

Fetches detailed profile data for a Vector trader by username.

fetch_token_data

Fetches trending Solana tokens. No parameters; returns tokens with 5-minute activity.

fetch_token_broadcasts

Fetches broadcasts for a specific Vector token by token_id.