Crawl4AI Plus AI Tools icon

Crawl4AI Plus AI Tools

Expose Crawl4AI web crawling and extraction as AI tools for AI Agent and MCP Trigger

Crawl4AI Plus AI Tools

AI-generated

Summary

Provide a unified AI tool interface that exposes all Crawl4AI web crawling and data extraction operations, selectable by the 'operation' parameter. Accepts all valid operations as input, validates parameters and credentials, executes the specified Crawl4AI operation, and returns structured JSON results including any extracted data or errors.

Inputs

  • operation (required) — The Crawl4AI operation to perform. Allowed values: crawl, askQuestion, extractWithLlm, extractWithCss, extractSeo, discoverLinks, healthCheck.
  • url — The full URL to crawl or extract data from (depending on operation). Must include protocol like https://.
  • question — For askQuestion operation: the specific question to ask about the page content.
  • instruction — For extractWithLlm operation: a text instruction describing what structured data to extract from the page.
  • schema — For extractWithLlm operation: optional JSON schema string defining the expected output data structure.
  • baseSelector — For extractWithCss operation: CSS selector for the repeating element to extract.
  • fields — For extractWithCss operation: JSON array string defining fields with name, selector, type, and optional attribute for extraction.
  • linkTypes — For discoverLinks operation: filter by link types. Allowed values: internal, external, both (default both).

Output shape

a single JSON response object with schemaVersion, success status, operation, resource, and either a result object with extracted or crawled data or an error object describing failure details

The output JSON varies by operation, including specific fields related to the crawl or extraction performed. Errors include standardized error types with guidance. The tool requires appropriate Crawl4AI credentials and may require LLM enabled credentials for LLM-based operations.

Examples

Example 1: Crawl a webpage

Set operation='crawl' and provide 'url'. Optional flags control inclusion of links and images and caching behavior. The node performs a fresh crawl or cached retrieval accordingly and returns page content as markdown with metadata and optional media and links.

Example 2: Ask an LLM question about a webpage

Set operation='askQuestion', provide 'url' and a specific 'question'. Requires LLM enabled credentials. The node crawls the page and queries the LLM to generate an answer with supporting details and source quotes.

Example 3: Extract structured data with LLM

Set operation='extractWithLlm' with 'url' and 'instruction' describing extraction needs. Optionally include 'schema' JSON string. Requires LLM enabled credentials. Returns extracted structured data as per the instruction/schema.

Links

Discussion