@pluckai/mcp

The Pluck MCP server: give any MCP client — Claude Code, Claude Desktop, Cursor, or your own agent — the ability to read the live web as clean markdown, crawl a site, search, extract structured JSON, and watch pages for changes.

Add it to a client

Hosted, no install:

{
  "mcpServers": {
    "pluck": {
      "url": "https://pluck-mcp.procd.cc/mcp",
      "headers": { "Authorization": "Bearer $PLUCK_API_KEY" }
    }
  }
}

Or over stdio:

{
  "mcpServers": {
    "pluck": {
      "command": "npx",
      "args": ["-y", "@pluckai/mcp"],
      "env": { "PLUCK_API_KEY": "pk_live_…" }
    }
  }
}

Get a key at pluck.procd.cc. New accounts start with 1,000 free credits.

Tools

| Tool | What the model gets | | --- | --- | | pluck_scrape | One page as markdown, HTML, links or images | | pluck_crawl | A whole site, with depth and path rules | | pluck_search | Web, news or image results, optionally already read | | pluck_map | Every URL on a site | | pluck_extract | JSON matching a schema you describe | | pluck_parse | PDFs and documents as markdown | | pluck_brand | Logos, colours, fonts and socials for a domain | | pluck_styleguide | Colours, type and spacing from a rendered page | | pluck_monitor | Watch a page and report what changed |

Point it somewhere else

PLUCK_API_URL (or PLUCK_API_INTERNAL_URL) sends the server at your own Pluck instance, so a self-hosted deployment works exactly the same:

{
  "env": { "PLUCK_API_KEY": "pk_live_…", "PLUCK_API_URL": "http://localhost:8080" }
}

Pluck is open source; the self-hosting guide has the compose file.

License

AGPL-3.0-only.