Mojeek AI Summary Scraper

What does Mojeek AI Summary Scraper do?

Mojeek AI Summary Scraper extracts public AI-generated answers from Mojeek Search. It is a structured Mojeek AI Summary API alternative for one question or a bounded list of questions. Each accepted dataset item contains the answer in plain text, sanitized HTML, and Markdown, plus the citations and public organic links shown with it.

The Actor does not solve CAPTCHAs, access private information, or fabricate a result when Mojeek does not provide a summary.

Why use Mojeek AI Summary Scraper?

• Collect privacy-search research answers with stable JSON fields.
• Preserve the query order and the source links behind inline citations.
• Run one query with query or a batch of up to 25 unique values with queries.
• Keep runs bounded through maxItems, low browser concurrency, strict timeouts, and retries.
• Use coherent Firefox fingerprints, persistent sessions, and optional Apify Proxy routing.
• Schedule runs and consume results through the Apify API, webhooks, and integrations.

The Actor validates HTTP status and content type before parsing, reads the rendered page before trusting the status code (Mojeek serves its ALTCHA challenge with HTTP 200), and evaluates Mojeek's robots.txt once per run. CAPTCHA, challenge, and access-denied pages fail closed with a self-describing OUTPUT record and are never written to the dataset. Results are buffered until the bounded crawl succeeds, so a failed batch does not leave misleading partial data.

What data can Mojeek AI Summary Scraper extract?

Field	Type	Description
id	string	Stable ID derived from query, language, and region
aiSummary	string	Clean plain-text answer
aiSummaryHtml	string	Sanitized semantic HTML with citation markers
aiSummaryMarkdown	string	Markdown representation of the answer
sources	array	Cited public source records with title, URL, and domain
inlineReferences	array	Citation positions, occurrences, and text offsets
organicResults	array	Supporting Mojeek organic results shown on the page
summaryTitle	string	Title displayed by Mojeek
disclaimer	string	Source disclaimer when displayed
queryPosition	integer	Position of the originating query
searchMetadata	object	Engine, request type, locale, and query position
scrapedAt	string	ISO-8601 extraction timestamp

Unavailable optional fields are omitted. Count fields can legitimately be zero. The Actor does not publish an invented model name or an estimated token count because Mojeek does not expose those values on the page.

How to scrape Mojeek AI summaries

1. Open the Actor input tab.
2. Enter a single question in Single search query, or add several questions under Multiple search queries.
3. Select a two-letter language and optional region.
4. Leave the default direct connection: Mojeek gates the /search route itself, so a proxy change does not remove the challenge.
5. Set Maximum summaries and Maximum concurrency for the size and sensitivity of the run.
6. Click Start and inspect the dataset after the run succeeds.

How much will it cost to scrape Mojeek?

This Actor uses Apify compute and, when enabled, proxy traffic. A single summary usually needs one short browser navigation. Residential Proxy traffic can cost more than direct or datacenter traffic, so use small limits and review the usage estimate for your Apify plan in Console. The Actor does not change Actor pricing or your subscription.

Input

See the input tab for every setting. At least one non-empty query or queries value is required at runtime.

• query: single-query mode.
• queries: multiple-query mode; duplicates are removed in first-seen order.
• maxItems: limits summaries and therefore limits queries processed.
• language and region: locale controls.
• maxConcurrency: 1–3 browser pages.
• maxRequestRetries: retries temporary failures only.
• navigationTimeoutSecs, aiTimeoutSecs, and requestHandlerTimeoutSecs: bounded waits.
• proxyConfiguration: no proxy, Apify Proxy, or user-provided proxy URLs.

GOOGLE_SERP cannot be used because that proxy group is limited to supported Google Search and Shopping requests, not Mojeek.

The complete input schema, including types, defaults, and bounds, is in .actor/input_schema.json.

Output

You can download the dataset in JSON, HTML, CSV, Excel, XML, or RSS formats and access it through the Apify API.

{
  "id": "mojeek-ai-summary:<stable-hash>",
  "type": "ai_summary",
  "globalPosition": 1,
  "queryPosition": 1,
  "query": "what is privacy-respecting search",
  "provider": "Mojeek",
  "summaryTitle": "Summary",
  "aiSummary": "<public summary text>",
  "aiSummaryLength": 344,
  "aiSummaryWordCount": 52,
  "sourcesCount": 2,
  "organicResultsCount": 10,
  "language": "en",
  "region": "GB",
  "scrapedAt": "2026-08-29T00:00:00.000Z"
}

The example illustrates the shape only; the Actor always stores live public data returned by Mojeek.

The complete output field schema is in .actor/dataset_schema.json.

Tips and advanced options

• Start with concurrency 1 for one query and 2 for a small batch.
• Keep media blocking enabled for faster runs; it does not remove summary text or source URLs.
• Direct access is the default, matching the Mojeek News Actor. Mojeek serves its ALTCHA challenge on /search regardless of proxy group or country, so proxy rotation is not a workaround.
• A valid search without an AI summary completes without an error record. If no requested query has a summary, the dataset is empty.
• Optional arrays are omitted when empty instead of being stored as empty placeholders.

FAQ, disclaimers, and support

Why did a run fail with a CAPTCHA or challenge?

Mojeek returned a CAPTCHA, challenge, or access-denied page for its /search route, which is also the only route that carries an AI summary (?fmt=summary). The Actor fails closed and does not retry or solve that page. Proxy changes do not clear the current Mojeek gate; wait until the service serves public results again before running it. A challenge page is never stored as a result.

The run's OUTPUT key-value record makes the refusal self-describing, for example:

{
  "status": "REFUSED",
  "errorCode": "TARGET_VERIFICATION",
  "httpStatus": 200,
  "pageUrl": "https://www.mojeek.com/search?q=what+is+privacy-respecting+search&fmt=summary&region=GB",
  "pageTitle": "Captcha",
  "pageHeading": "Verification required",
  "pageExcerpt": "Verification required Please complete the challenge to continue. I'm not a robot Protected by ALTCHA",
  "robotsMatchedRule": "Disallow: /search",
  "robotsSearchDisallowed": true
}

https://www.mojeek.com/robots.txt publishes User-agent: * / Disallow: /search, so the only query-driven Mojeek surface is off-limits to automated clients independently of the challenge. The Actor records that rule instead of hiding it.

Why is there no dataset item for a query?

Mojeek may return ordinary organic results without generating an AI summary. That is a legitimate no-summary outcome, not an error record.

Is the output safe to render?

aiSummaryHtml is reduced to a small allowlist of semantic tags, scripts and active content are removed, and scraped text is treated as untrusted data. Consumers should still follow their own output-encoding rules.

This Actor extracts only publicly displayed Mojeek content. Results can link to third-party pages that may contain personal data. Use the Actor only for a lawful purpose, follow applicable terms and robots guidance, and consult legal counsel if you are unsure whether your use is legitimate.

Use the Actor Issues tab for support and the API tab for programmatic examples.