Web Scraping & Search: Crawling, Tavily, Exa

Web scraping and search APIs let you ingest competitor sites, documentation, and search results. Graphlit crawls websites and integrates with Tavily and Exa search APIs.

What You'll Learn

Web crawling configuration
Site mapping and link following
Domain and path filtering
Tavily search integration
Exa search integration
Web-to-markdown conversion

Part 1: Web Crawling

import { FeedTypes } from 'graphlit-client/dist/generated/graphql-types';

// Crawl documentation site
const webCrawl = await graphlit.createFeed({
  name: 'Documentation Crawler',
  type: FeedTypes.Web,
  web: {
    uri: 'https://docs.example.com',
    readLimit: 500,
    allowedPaths: ['^/docs/.*'],      // Regex: only crawl /docs/* pages
    excludedPaths: ['^/api/.*', '^/archive/.*']  // Regex: skip these sections
  }
});

What happens:

Starts at uri
Extracts page content → markdown
Follows links matching allowedPaths patterns
Continues until readLimit pages reached

Part 2: Sitemap Crawling

// Crawl from sitemap.xml using Web feed with includeFiles
const sitemapCrawl = await graphlit.createFeed({
  name: 'Sitemap Crawler',
  type: FeedTypes.Web,
  web: {
    uri: 'https://example.com/sitemap.xml',
    includeFiles: true,  // Include files referenced in sitemap
    readLimit: 1000
  }
});

Benefits:

Faster than link following
Gets all pages from sitemap
Respects site structure

Part 3: Tavily Search

import { FeedTypes, SearchServiceTypes } from 'graphlit-client/dist/generated/graphql-types';

// Ingest Tavily search results
const tavilyFeed = await graphlit.createFeed({
  name: 'Tavily Search',
  type: FeedTypes.Search,
  search: {
    type: SearchServiceTypes.Tavily,
    text: 'AI agent frameworks',
    readLimit: 50
  }
});

Use cases:

Market research
Competitive intelligence
Trend analysis

Part 4: Exa Search

import { FeedTypes, SearchServiceTypes } from 'graphlit-client/dist/generated/graphql-types';

// Ingest Exa search results
const exaFeed = await graphlit.createFeed({
  name: 'Exa Search',
  type: FeedTypes.Search,
  search: {
    type: SearchServiceTypes.Exa,
    text: 'machine learning papers',
    readLimit: 100
  }
});

Alternative: Direct search without feed

// One-time search (not continuous monitoring)
const results = await graphlit.searchWeb(
  'machine learning papers',
  SearchServiceTypes.Exa,
  100  // limit
);

Production Patterns

Competitive Intelligence

// Crawl competitor docs
const competitors = [
  'https://competitor1.com/docs',
  'https://competitor2.com/docs',
  'https://competitor3.com/docs'
];

for (const url of competitors) {
  await graphlit.createFeed({
    name: `Competitor: ${url}`,
    type: FeedTypes.Web,
    web: {
      uri: url,
      readLimit: 200,
      allowedPaths: ['^/docs/.*']  // Stay within docs section
    }
  });
}

// Search across all competitors
const features = await graphlit.queryContents({
  filter: {
    search: 'pricing features'
  }
});

Documentation Monitoring

import { TimedPolicyRecurrenceTypes } from 'graphlit-client/dist/generated/graphql-types';

// Monitor docs for changes
const docsCrawl = await graphlit.createFeed({
  name: 'Docs Monitor',
  type: FeedTypes.Web,
  web: {
    uri: 'https://docs.example.com',
    readLimit: 500
  },
  schedulePolicy: {
    recurrenceType: TimedPolicyRecurrenceTypes.Repeat,
    repeatInterval: 'P1D'  // ISO 8601 duration: 1 day
  }
});

Related Guides

Data Connectors - All connector types
Content Ingestion - Ingestion basics