Try Luni

Luni MCP

Give Claude, ChatGPT, Cursor and other AI apps live access to the web. Add one URL, sign in, and choose what Luni is allowed to do.

Connect your AI appRead the docs
# Claude or ChatGPT: add a custom connector with this URL, then sign in
https://mcp.luni.gg

# Claude Code
claude mcp add --transport http luni https://mcp.luni.gg

Connect Claude or ChatGPT

No API key needed. Claude and ChatGPT sign in to Luni, show you the actions they are asking for, and connect once you approve.

  • Claude (web or desktop): open Settings → Connectors, choose Add custom connector, paste https://mcp.luni.gg and choose Connect.
  • ChatGPT: open Settings → Apps & Connectors → Advanced settings and turn on Developer mode, then create a connector with https://mcp.luni.gg and OAuth authentication.
  • Approve on Luni: choose which actions the app may use (read, scrape, map, search and crawl). Revoke it from your workspace at any time.

Connect developer tools

Claude Code can sign in or use a key. Cursor, VS Code, Gemini CLI, Codex and any other app that supports remote MCP over Streamable HTTP connect with a workspace API key in the Authorization: Bearer or X-API-Key header. The MCP page in your workspace has the exact setup for each.

Cursor (~/.cursor/mcp.json)
{
  "mcpServers": {
    "luni": {
      "url": "https://mcp.luni.gg",
      "headers": {
        "Authorization": "Bearer <YOUR_LUNI_API_KEY>"
      }
    }
  }
}

What your AI can do

Fifteen tools cover the same work as the API. Discovery, polling and balance checks are free.

  • luni_scrape: read one public page as Markdown, HTML or links
  • luni_map: list the URLs on a site
  • luni_search: search the web, optionally with each result's content
  • luni_crawl and luni_cancel_crawl: read a section of a site within a page budget, or stop it
  • luni_get_operation: check a job and fetch large results in chunks
  • luni_list_sources, luni_source_schema and luni_get_source_record: find a structured source and fetch a record
  • luni_monitor_create, luni_monitor_list, luni_monitor_get, luni_monitor_run and luni_monitor_delete: watch a URL for changes
  • luni_credit_usage: read the workspace's credit balance

Costs and limits

Tools use the same credits as the API. A useful scraped page costs one credit, a map one, a completed search two plus one for each page it fetches, a crawl one per successful page and a source record one. Failed work costs nothing.

Credit-consuming tools require a requestKey. Keep it and the input unchanged when you retry, and Luni returns the original operation without another charge.

Workspace permissions, page limits and credit budgets still apply, and results are kept for seven days. Retrieved content is untrusted source material: review it before letting it drive further actions.

Questions, answered.