wigolo

Install

npx skillfoxx add mcp/wigolo

claude mcp add --transport stdio wigolo -- npx -y wigolo

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

code --add-mcp '{"name":"wigolo","type":"stdio","command":"npx","args":["-y","wigolo"]}'

{
  "servers": {
    "wigolo": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

codex mcp add wigolo -- npx -y wigolo

[mcp_servers.wigolo]
command = "npx"
args = ["-y", "wigolo"]

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcp": {
    "wigolo": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "context_servers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

{
  "mcpServers": {
    "wigolo": {
      "command": "npx",
      "args": [
        "-y",
        "wigolo"
      ]
    }
  }
}

npx wigolo init --agents=claude-code,cursor

This is third-party code. Review the repository files before installing.

What it does

wigolo gives the agent one surface for everything web: search, fetch, crawl, extract, cache, find-similar, research and autonomous gather loops. It runs where the agent runs: as an MCP server next to a coding agent, as a REST or MCP endpoint for self-hosted agents or embedded through an SDK. The core tools need no keys, nothing it touches leaves the ~/.wigolo directory, and the bill does not grow with the number of queries. The init command sets up the local engine, downloads a browser engine and models and can write the MCP config for the agents you name. For synthesized answers in research and agent you can add a free model key; without it you get raw material back.

Who it is for. For developers and agent owners who want private web access without paid search APIs or the cloud.

Example request

Search the web for recent material on the topic, synthesize it and cite the sources

Limitations

The project is in public beta. It needs Node 20 or newer and about 1.5 GB of disk for the browser engine and models. Search, fetch, crawl, extract, cache and find-similar are keyless, while research, agent and the answer mode use a model and need a free key, otherwise they return a raw brief. Data stays in the ~/.wigolo directory.

How to disable. Run npx wigolo config --uninstall --yes and remove the wigolo entry from the MCP client config.

MCP

Transport

stdio, http

Authentication

not required

Security check

• Goes to the network, fetches and crawls pages
• Downloads a browser engine and models to the machine

README in short

The README describes wigolo as a local web engine for agents: search, fetch, crawl, extract, cache, find-similar, research and gather loops through one interface. It runs as an MCP server next to the agent, as a REST or MCP endpoint and through SDKs for Python, TypeScript, LangChain, CrewAI, LlamaIndex, the Vercel AI SDK and n8n. The init command sets up the engine and writes config for the chosen agents, and doctor checks health. Core tools are keyless; synthesized answers need a model key. The package is on npm, AGPL-3.0 licensed.

Related