Crawlbase MCP

When to choose Crawlbase MCP

Connect to Crawlbase MCP

Choose a connection profile below and replace credential placeholders in your own MCP client. The transport describes the MCP connection.

{
  "mcpServers": {
    "crawlbase-stdio": {
      "command": "npx",
      "args": [
        "-y",
        "@crawlbase/mcp"
      ],
      "env": {
        "CRAWLBASE_TOKEN": "YOUR_CRAWLBASE_TOKEN",
        "CRAWLBASE_JS_TOKEN": "YOUR_CRAWLBASE_JS_TOKEN"
      }
    }
  }
}

Fetch, inspect and reuse a page

A suggested sequence using the documented tools. Check the connected server's tool schema before making a call.

1. Select the output

   Use Markdown for reading or HTML for markup inspection. Enable storage when the page needs to be retrieved again.

   Tools: crawl, crawl_markdown

   Sources: Crawlbase MCP tools and setup

2. Check rendering when needed

   Take a screenshot for a visual check. Preserve the returned request identifier and token_type with your notes.

   Tools: crawl_screenshot

   Sources: Crawlbase MCP tools and setup, Crawlbase Cloud Storage reference

3. Retrieve the saved version

   Locate the RID and fetch the stored record. Select the JavaScript token scope for a page originally captured with that token.

   Tools: storage_list, storage_get

   Sources: Crawlbase Cloud Storage reference

Selected tools & example tasks

• crawl

  Retrieve HTML for a URL.

  Example task: Inspect markup or save a page.

  Sources: Crawlbase MCP tools and setup
• crawl_markdown

  Retrieve page content as Markdown.

  Example task: Read a page in the assistant.

  Sources: Crawlbase MCP tools and setup
• crawl_screenshot

  Return a rendered PNG image.

  Example task: Check the visible page.

  Sources: Crawlbase MCP tools and setup
• storage_get

  Retrieve a stored page by RID or URL.

  Example task: Reuse a recorded crawl.

  Sources: Crawlbase MCP tools and setup
• storage_list

  List stored request identifiers with pagination.

  Example task: Locate saved results.

  Sources: Crawlbase MCP tools and setup

Troubleshoot the connection

Limits & undocumented behavior

• The local profile needs separate Normal and JavaScript tokens for their respective operations.
• Storage belongs to the token used for the crawl.
• Documentation was reviewed; authenticated requests were not tested.

Sources & review history

This listing is documentation-confirmed. We reviewed the primary sources below; this is not a successful live connection test.

Record updated 2026-09-27. Review due 2026-10-27.

1. Crawlbase MCP tools and setupChecked 2026-09-27 · Package, tokens, tool names and screenshot storage behavior.
2. Official Crawlbase MCP repositoryChecked 2026-09-27 · Local stdio configuration and authentication.
3. Crawlbase Cloud Storage referenceChecked 2026-09-27 · Persistence, token scope and retrieval formats.
4. Crawlbase pricingChecked 2026-09-27 · Request credits and shared product billing.
5. Crawlbase published MCP package metadataChecked 2026-09-27 · Package name, Node.js 18 runtime requirement and MIT license.

Spotted a change? Share the source and the affected setup step in Site Feedback.

Alternatives to Crawlbase MCP

These integrations share documented capabilities with Crawlbase MCP. Compare their setup and product requirements for your task.