Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy

Loading page…

Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy

Loading page…

Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy
Customer Stories

SpoilerScroll extracts article text to identify what is new

Written by
Sanjit Juneja
Published
Oct 18, 2026

Extract article text and separate new information from material the reader has already seen.

Developer project · Proposed Act workflow.

The build.

SpoilerScroll is a Chrome extension with a TypeScript backend. It calls Darwin Search when local text extraction is insufficient. The repository records successful discovery, but no executable provider route or live extraction through Darwin. Its optional model service is configurable; no vendor is assumed.

View implementationRead the project notes

The brief

Readers often encounter the same information across several articles. SpoilerScroll compares reading material and uses text extraction to help identify what is new.

Finding the right capabilities

Darwin finds extraction capabilities for the selected sources. Coverage matters: missing text limits what a summary can establish.

Coordinating the work

An extraction agent could return permitted text with source references. SpoilerScroll’s own summarizer would compare it with the reader’s existing context.

What comes back

New information separated from repeated coverage, with source links and extraction gaps available to inspect.

The plan, step by step

  1. 1

    Find an extractor.

    a specialist comparing documents at her desk.

    Search. Find text-extraction capabilities that support the selected article sources.

    Act. Check access and coverage limits before requesting article text.

    Outcome. An extraction plan showing supported sources and expected coverage gaps.

  2. 2

    Keep the source with the text.

    Search. Use the chosen extractor for the approved sources.

    Act. Request text with article references and a record of missing sections or extraction failures.

    Outcome. Extracted text with source links and coverage notes for summarization.

  3. 3

    Identify what is new.

    Search. SpoilerScroll’s own summarization compares the extracted articles with the reader’s existing context.

    Act. Produce the comparison locally; ask the extractor for clarification if missing text prevents a conclusion.

    Outcome. New points, repeated topics, source links, and limits caused by incomplete extraction.

Reading comparison.

Follow each new point back to extracted source text and check where coverage is incomplete.

Extraction plan

Deliverable: An extraction plan showing supported sources and expected coverage gaps.

Review criteria: Keep inaccessible and unsupported sources visible. Do not bypass access restrictions or infer missing article text.

Extracted text

Deliverable: Extracted text with source links and coverage notes for summarization.

Review criteria: Mark partial extraction as partial. Do not replace unavailable sections with generated content.

Reading comparison

Deliverable: New points, repeated topics, source links, and limits caused by incomplete extraction.

Review criteria: Base the comparison only on available text. Keep repeated coverage and genuinely new points separate.

Similar articles

All customer stories
TechnologyCharlotte

Charlotte turns meeting notes into follow-up tasks with Python and Slack.

TechnologyPrecedent

Precedent turns drive-health alerts into replacement searches.

TechnologyAxolotl

Axolotl finds another path when an agent workflow fails with TypeScript.

Darwin

ProductsNavigatorOriginalsPricing
ResourcesFor DevelopersFor BusinessCase Studies
CompanyNewsAboutCareers
LegalTerms of ServicePrivacy PolicyAcceptable Use Policy

© 2026 Darwin Technologies, Inc.