Mojeek AI Summary Scraper

Extract Mojeek's experimental LLM-generated AI summaries for any search query, along with the organic search results that serve as citation sources.

What This Actor Does

Mojeek displays AI-generated summaries (powered by Large Language Models) at the top of search results for many queries. This actor:

• Navigates to https://www.mojeek.com/search?q=QUERY&fmt=summary
• Extracts the full AI summary text (both plain and Markdown formats)
• Records which source positions are cited inline (e.g. [1][3][5])
• Extracts all 8 organic search results shown alongside the summary (these are the citation sources)
• Includes the disclaimer text and a unique session ID when Mojeek supplies them
• Detects CAPTCHA/block pages before extraction and never stores them as data

Input

Field	Type	Default	Description
queries	array<string>	required	List of search queries to process
query	string	—	Single query shortcut (alternative to queries)
language	string	"en"	Language code (e.g. en, fr, de)
region	string	""	Country code for regional results (e.g. GB, US)
proxyConfiguration	object	—	Apify proxy settings (recommended for cloud runs)

Example Input

{
  "queries": ["artificial intelligence", "climate change", "quantum computing"],
  "language": "en",
  "region": ""
}

Output

Each query with a real Mojeek AI summary produces one dataset item. Queries with no summary or a blocked response are counted in OUTPUT_SUMMARY and are not written as placeholder/error records. A run with zero real summaries fails.

Field	Type	Description
query	string	The search query used
hasSummary	boolean	Always true for accepted dataset records
aiSummary	string	Plain text summary (markdown stripped)
aiSummaryMarkdown	string	Raw summary with **bold** and [N] citation markers
aiSummaryLength	number	Length of the markdown summary in characters
summaryId	string	Unique session ID for this summary (data-suid)
disclaimer	string	Mojeek's disclaimer about AI accuracy
citedSourcePositions	array<number>	Positions cited inline, e.g. [1, 3, 5]
citedSourcesCount	number	Number of unique citation positions
organicResults	array<object>	The 8 source pages listed alongside the summary
organicResultsCount	number	Number of organic results (usually 8)
searchUrl	string	Final URL used for this query
language	string	Language used
region	string	Region used, omitted when not configured
sourceDomain	string	mojeek.com
extractionMethod	string	DOM extraction strategy
scrapedAt	string	ISO timestamp

organicResults Item Fields

Field	Type	Description
position	number	Citation position (1–8)
title	string	Page title
url	string	Full URL
displayedUrl	string	Displayed URL under the title
snippet	string	Short description (when available)

Example Output

{
  "query": "artificial intelligence",
  "hasSummary": true,
  "aiSummary": "Artificial Intelligence (AI) refers to the capability of computational systems to perform tasks typically associated with human intelligence [1][3][5]...",
  "aiSummaryMarkdown": "**Artificial Intelligence (AI)** refers to the capability... [1][3][5]...",
  "aiSummaryLength": 772,
  "summaryId": "78c731e97d320257",
  "disclaimer": "Summaries are experimental and generated by large language models...",
  "citedSourcePositions": [1, 2, 3, 4, 5, 7, 8],
  "citedSourcesCount": 7,
  "organicResults": [
    {
      "position": 1,
      "title": "Artificial intelligence - Wikipedia",
      "url": "https://en.wikipedia.org/wiki/Artificial_Intelligence",
      "displayedUrl": "https://en.wikipedia.org › wiki › Artificial_Intelligence",
      "snippet": "Public source snippet"
    }
  ],
  "organicResultsCount": 8,
  "searchUrl": "https://www.mojeek.com/search?q=artificial+intelligence&fmt=summary&kalid=78c731e97d320257",
  "language": "en",
  "region": null,
  "scrapedAt": "2026-05-12T17:21:43.011Z"
}

Dataset Views

Two views are available in the Apify console:

• AI Summary Results Overview — Compact table: query, hasSummary, summary length, cited sources count, organic count, scrapedAt
• Full AI Summaries — Extended table: includes the full aiSummary, aiSummaryMarkdown, disclaimer, citedSourcePositions, and searchUrl

Notes

• Experimental feature: Mojeek AI summaries are not always available. Queries without a real summary produce no dataset item and are counted in the run summary.
• No pagination: Each query produces exactly one summary (there is no multi-page summary).
• Source details: Inline citation details (span.details) are dynamically loaded by Mojeek on user click, so citedSourcePositions tracks which organic results are referenced rather than fetching dynamic data.
• Proxy: Start without a proxy. If Mojeek returns its public CAPTCHA page, configure an Apify Proxy group available to your account; the Actor fails closed if access remains blocked.
• Browser: Uses Firefox (Playwright) for JavaScript rendering.

Bounded QA run

npm ci
apify run --purge --input-file qa-inputs/local-search.json

Cloud validation uses qa-inputs/cloud-validation.json with one query and a bounded proxy configuration. The Actor supports at most 20 unique queries per run and closes its browser/context in all success and failure paths.

Use Cases

• AI research: Compare how LLMs summarize the same topic across queries
• Content monitoring: Track what sources Mojeek's AI cites for specific topics
• SEO analysis: See which pages are cited as authoritative sources by Mojeek AI
• Query benchmarking: Evaluate summary quality and length across different queries