Content Brief Source Collector

Collect traceable evidence for content briefs from supplied source pages and structured search-result records. The Actor extracts page titles, heading patterns, visible questions, content-format signals, numeric claim candidates, and bounded source context. Every result preserves its source URL so writers and editors can verify evidence before using it.

Use it for research bundles, brief preparation, writer handoffs, competitor-source inventories, and question or heading analysis. The Actor identifies candidate evidence; it does not judge source authority, create unsupported facts, or turn a number into a verified claim automatically.

What the Actor extracts

• the accepted source title and target query;
• visible H1–H6 headings in source order;
• headings, summaries, or definition terms ending in a question mark;
• tables, lists, images or figures, video or embeds, and FAQ markup;
• numeric sentences from visible paragraphs, list items, and blockquotes;
• surrounding source context and a transparent evidence-priority score.

claimExtractionMode controls numeric candidate selection. off returns a source-level evidence record without a numeric claim. standard requires both a number and research-oriented wording. aggressive keeps any visible paragraph, list item, or blockquote containing a number. None of these modes verifies that the text is accurate.

Input

Field	Purpose	Default / cap
targetQueries	Brief topics associated with accepted sources	one bounded default query
sourceUrls	Source pages to retrieve	empty
serpResults	Structured query/result records with URL, title, snippet, text, or HTML	empty
maxSources	Maximum sources processed	10 default, 25 maximum
maxEvidenceRecords	Maximum numeric candidates per source	25 default, 100 maximum
claimExtractionMode	off, standard, or aggressive	standard
includeQuestions	Extract visible question signals	true
includeFormatSignals	Detect supported content formats	true
htmlInputs	Captured source records with sourceUrl and HTML	empty
allowedDomains	Optional retrieval hostname allowlist	empty
maxPages	Maximum pages retrieved	1 default, 25 maximum
includeEvidence	Preserve bounded evidence excerpts	true
includeRawSnapshots	Store accepted raw source HTML	false
maxChargeUsd	Stop before the next event exceeds this amount	35 USD maximum

Example:

{
  "targetQueries": ["project management software"],
  "sourceUrls": [
    "https://research.example/workflow-study",
    "https://publisher.example/project-guide"
  ],
  "maxSources": 2,
  "maxEvidenceRecords": 20,
  "claimExtractionMode": "standard",
  "includeQuestions": true,
  "includeFormatSignals": true,
  "allowedDomains": ["research.example", "publisher.example"],
  "maxChargeUsd": 5
}

Structured serpResults can be used when a trusted upstream workflow already captured results. Those records are analyzed directly and do not require a search-engine request. The Store default has no source evidence and returns a nonempty source_required diagnostic.

Dataset output

Each row includes common provenance plus:

• briefTopic for the associated target query;
• sourceTitle for the observed page title;
• headingSignals and questionSignals as arrays of visible text;
• claimText for a numeric candidate, or null for a source-level record;
• claimContext for the bounded supporting excerpt;
• contentFormat for the detected format list;
• evidencePriority as a transparent score based on observed claims, headings, questions, and formats;
• issueCode, evidenceSnippet, and recommendation for audit use.

Evidence priority is a sorting aid, not an authority or truth score. Review publication date, methodology, authorship, original citations, and source reputation before putting a claim into published content.

Reports

When includeReport is enabled, key-value storage receives a Markdown summary, CSV records, issue JSON, bounded evidence excerpts, and run metadata. Raw HTML is written only when explicitly requested.

Pay-per-event pricing

Event	Price (USD)	When charged
actor-start	0.0350	after successful startup
source-page-processed	0.0115	once for each accepted source page
evidence-record-extracted	0.0065	before each paid evidence record
brief-section-generated	0.0500	before report section generation
premium-brief-generated	0.7500	before the premium brief artifacts are stored

Apify platform usage pass-through is off. Higher-tier Store discounts are reserved in the pricing policy. maxChargeUsd is evaluated before every event, and no related dataset or report output is written after a rejected, partial, or over-budget charge.

Safety and responsible research

Only HTTP and HTTPS source URLs are accepted. Loopback and private IPv4 destinations are rejected, optional hostname allowlists narrow retrieval, and source counts, page counts, response size, and timeouts remain bounded. Discovered links are not crawled. Access-denied and human-verification pages return diagnostics.

The Actor works only with text present in accepted responses or supplied structured records. It does not retrieve private content, validate legal rights to reuse a source, or guarantee factual accuracy. Cite original research where possible and preserve the source URL with every downstream brief record.