Skip to main content

Discovering MCP Tools

Nexus is fully MCP-compatible, which means your AI agents can be equipped with thousands of tools from the Smithery.ai ecosystem and beyond. This includes tools for APIs, data processing, visualization, scraping, automation, and much more.

To build powerful agents, you first need to discover which tools are available, explore what they do, and understand how to configure them.

Get Built-in Nexus Tools​

GET /mcp/client/nexus

This returns a list of tools that come bundled with Nexus by default. These require no external setup and are ready to use.

Response example:

{
  "status": "ok",
  "data": {
    "tools": [
      {
        "tool_name": "find_metadata",
        "description": "Searches column-level metadata across all connected data sources.",
        "tool_server_id": "nexus",
        "config": {}
      },
      {
        "tool_name": "execute_sql",
        "description": "Executes Nexus SQL queries across structured datasets, databases and API endpoints.",
        "tool_server_id": "nexus",
        "config": {}
      },
      {
        "tool_name": "create_graph",
        "description": "Generates and returns charts from tabular result data using matplotlib.",
        "tool_server_id": "nexus",
        "config": {}
      }
    ]
  }
}

These tools form the backbone of most query + visualization workflows in Nexus.

Search MCP Tool Servers​

GET /mcp/client/search?q={query}&page={page}

Use this to search the global MCP ecosystem for servers offering tools related to your query. This is the best way to find capabilities beyond what's bundled in Nexus.

  • q — a text search query (e.g., "weather", "salesforce", "image").
  • page — optional; defaults to 1.

Example request:

GET /mcp/client/search?q=image

Response example:

{
  "status": "ok",
  "data": {
    "servers": [
      {
        "qualifiedName": "@PawNzZi/image-server",
        "displayName": "text2image",
        "description": "Transform text prompts into stunning images effortlessly…",
        "createdAt": "2025-04-08T04:04:08.356Z",
        "useCount": 120,
        "homepage": "https://smithery.ai/server/@PawNzZi/image-server"
      }
    ]
  }
}

Get MCP Server Details​

GET /mcp/client/server?q={qualified_name}

After finding a promising server via search, use this endpoint to get its full details, including:

  • The tools it provides
  • Input schemas and config options
  • Deployment URLs
  • Security scan status

Example request:

GET /mcp/client/server?q=@PawNzZi/image-server

Response example:

{
  "status": "ok",
  "data": {
    "server": {
      "qualifiedName": "@PawNzZi/image-server",
      "displayName": "text2image",
      "remote": true,
      "iconUrl": "https://icons.duckduckgo.com/ip3/github.com.ico",
      "connections": [
        {
          "type": "http",
          "deploymentUrl": "https://server.smithery.ai/@PawNzZi/image-server/mcp",
          "configSchema": {}
        }
      ],
      "security": {
        "scanPassed": true
      },
      "tools": [
        {
          "name": "image_generation",
          "description": "Image generation assistant, please imagine and describe a complete picture…",
          "inputSchema": {
            "type": "object",
            "required": ["image_prompt"],
            "properties": {
              "width": { "type": "integer", "default": 1024 },
              "height": { "type": "integer", "default": 1024 },
              "image_prompt": { "type": "string" }
            }
          }
        }
      ]
    }
  }
}

How to Use This Data​

Once you've identified a tool you want to use:

  1. Get its qualified name (e.g., @PawNzZi/image-server)

  2. Get server details to retrieve its deploymentUrl, config schema, and list of tools.

  3. Add the tool to your agent via the POST /agent endpoint by specifying:

    • tool_name
    • tool_server_id (the qualified name)
    • Any required properties based on its input or config schema

Read more about how to configure agent and agent tools here.

Tools that require user-provided config (like API keys, usernames, or prompts) will list those in the properties field or input schema.