skim.page
Add skim.page/ before any article URL and the site builds a ready-made summarization prompt, then hands it to ChatGPT, Claude, Gemini, Perplexity or Grok. Four reading angles, no account, no installation, entirely free.
What is skim.page?
skim.page is a URL shortcut rather than a summarization engine. Type skim.page/ in front of any article address, as in skim.page/https://example.com/article, and the site assembles a ready-made summarization prompt for that article, then redirects the current tab to the AI assistant of your choice with the prompt already filled in. The summarizing itself is done entirely by that assistant and never by skim.page, a distinction the site is unusually careful to spell out.
Four reading angles, called summary styles, are built in, each with its own fixed prompt template. Summary, the default, asks for a one or two sentence TL;DR, three key takeaways and a line on why the article matters. ELI5 requests plain language, three to five simple bullets and one real-world analogy. Research produces a structured breakdown: main argument, three pieces of evidence, any major assumptions or limitations, and one open question. Investor reframes the piece as a market story, with a bull case, a bear case, a business implication and a metric or risk to watch.
Five assistants are supported. ChatGPT is the default, alongside Claude, Gemini, Perplexity and Grok. Each behaves differently on handoff: ChatGPT and Claude open with the prompt prefilled and wait, while Gemini, reached through Google's AI Mode search, along with Perplexity and Grok, may run it immediately. Claude requires an active session before the prompt can open at all.
Styles and assistants combine through one or two letter shortcuts, so skim.page/i/cl/ followed by an article URL opens the Investor angle in Claude. The same request can also be expressed as ordinary query parameters.
Technically the tool is deliberately thin. It is a React and TypeScript single-page application built with Vite and served from Vercel, with no backend, no user accounts, no database and no article scraping: everything happens in your browser. Its source code is published in a public GitHub repository, and a copy button hands you the raw prompt whenever a redirect fails.
What it does
- Turn any article URL into a ready-made AI summarization prompt
- Hand that prompt straight to ChatGPT, Claude, Gemini, Perplexity or Grok
- Switch between four reading angles: Summary, ELI5, Research and Investor
- Pin a style and an assistant into the URL itself with one or two letter shortcuts
- Preview the exact prompt before it is sent
- Copy the prompt to the clipboard when a redirect fails
- Bookmark the URL pattern for repeated one-step use
When to use skim.page / When not to
A quick filter to help you decide if skim.page is the right fit.
When to use skim.page
- Heavy news and long-read consumers who want a TL;DR before deciding whether to read in full
- Researchers and students who need a structured breakdown of an article's argument and evidence
- Analysts and investors reading market coverage who want a bull case, bear case and key risk
- Journalists, editors and communications staff doing fast daily monitoring across many sources
- People who already pay for ChatGPT, Claude, Gemini, Perplexity or Grok and want less friction
When not to use skim.page
- Anyone hoping to read paywalled articles, which the tool explicitly refuses to circumvent
- Users without access to any of the five supported assistants, since nothing works without one
- Teams needing an API, batch processing or automation, none of which exist here
- Organisations with procurement or compliance requirements, as no legal entity, terms of service or GDPR statement is published
- Anyone wanting saved summaries or a reading history, since the tool stores nothing and has no accounts
How to use skim.page
A typical end-to-end flow, from setup to results.
- Open any news article, blog post or long-read in your browser
- Click into the address bar and place the cursor before the article URL
- Type skim.page/ immediately in front of it, then press Enter
- Optionally insert a style shortcut first: e for ELI5, r for Research, i for Investor
- Optionally add an assistant shortcut: ge for Gemini, cl for Claude, px for Perplexity, gr for Grok
- Combine both if wanted, as in skim.page/i/cl/ followed by the article URL
- Let the tab redirect to the chosen assistant, where the prompt arrives prefilled
- Sign in to the assistant if prompted, which Claude in particular usually requires
- Send the prompt, or let Gemini, Perplexity or Grok run it automatically
- If the redirect fails, return to skim.page and use the copy button, then paste the prompt manually
Pros & Cons
Pros
- Almost zero friction: nothing to install, register for or configure
- Completely free, with no account, no paid tier and no usage limit
- Memorable URL pattern that works from any browser on any device
- Keeps you inside the assistant you already pay for instead of adding another subscription
- Four purpose-written angles beat a generic improvised prompt
- Unusually candid about its own limits, including paywalls and assistant dependence
- Minimal data surface and publicly readable source code
Cons
- Produces no summary of its own, so quality depends entirely on the chosen assistant
- Relies on the assistant being able to open the URL, which is the usual point of failure
- No legal entity, no terms of service, no address and no contact email are published anywhere
- No GDPR statement and no stated retention period
- Assistant handoff URLs can change without notice and break the tool, a limitation the author acknowledges
- A personal side project at version 0.1.0, with no licence declared and no commit since 11 May 2026
- No API, no batch processing, no saved history and no support channel beyond GitHub issues
Pricing & Plans
skim.page is entirely free. There is a permanent free offering with no paid tier, no subscription, no credit system and no usage cap, and no price is published anywhere on the site, in its JavaScript bundle or in its public source repository. The only indirect cost is access to one of the five supported AI assistants, some of which require a paid plan for their better models.
- Free — unrestricted access to all four summary styles and all five AI assistants
- with no account and no paid tier
Data, GDPR & hosting
A consolidated view of how skim.page handles your data.
GDPR overview
There is no mention of the GDPR anywhere on skim.page. The site publishes a short privacy page of three paragraphs, but it contains no reference to the General Data Protection Regulation, no legal basis for processing, no data subject rights such as access, rectification or erasure, and no retention period. No data protection officer and no Article 27 representative are named, and no data controller is identified, since the site discloses no legal entity at all. This absence should be read as silence rather than as a declaration of non-compliance: the operator neither claims compliance nor denies it. European users have no published contact point for exercising their rights, the only channel offered anywhere on the site being a public GitHub issue tracker.
Who owns the data?
skim.page publishes no terms and conditions, so no contractual clause assigns ownership of anything to anyone. The privacy page states only that the tool is client-only, with no user accounts, no database and no article scraping, and that prompts and article URLs are not intentionally stored. In practice there is little for the operator to own: the prompt is assembled in your own browser from a fixed template and immediately handed to a third-party assistant. Ownership questions therefore shift entirely to that assistant and to the publisher of the article you submit. No named legal entity stands behind the service, which leaves no identifiable counterparty for any ownership claim or dispute.
Reuse rights
No terms and conditions exist, so nothing on the site either grants or restricts reuse of what you generate. The prompt is plain text produced in your own browser from a fixed template plus the article URL you supplied, and the operator states no condition on what you do with it. Reuse of the resulting summary is governed instead by the AI assistant that produced it and by the copyright of the source article, neither of which skim.page controls, licenses or comments on anywhere.
Data retention & training
Hosting summary
skim.page names no hosting country, no region and no jurisdiction. Its privacy page refers only to "skim.page hosting" and to "server/CDN logs" in general terms, without identifying any provider, and there is no trust or security page, no subprocessor list and no data processing agreement. Because the tool is client-only, with no accounts and no database, there is no stored user data to locate in the first place: what reaches the infrastructure is the article URL required to serve the page, plus whatever ordinary server and CDN logging retains. Independently of any statement by the operator, technical observation places the site on Vercel in the United States: the domain resolves to 216.198.79.1, geolocated in the United States, and the public repository contains a Vercel configuration file and a vercel.app address. This is an observation rather than a commitment, and nothing prevents it from changing. Users in regulated environments should treat the hosting jurisdiction as undeclared.
Things to keep in mind
Risks and trade-offs to weigh before adopting skim.page.
- Your reading list becomes visible: every article URL you submit passes through third-party hosting and lands with an AI assistant, which may be sensitive for legal, medical or journalistic research
- No legal entity, terms of service, postal address or contact email is published, so there is no accountable counterparty and no route for a complaint
- Summaries can be confidently wrong: the assistant may not reach the article at all and can still produce plausible output, so verify before relying on it
- Habitual skimming can erode close reading, and a four-bullet TL;DR flattens nuance, caveats and disagreement that matter in contested subjects
- The Investor angle produces market framing that is not financial advice and rests on a single article read by a general-purpose model
- The tool is an unmaintained personal project at version 0.1.0 with no licence, and the assistant URLs it relies on can break without warning
- The domain previously hosted a different product, so older reviews, bookmarks or links referring to skim.page may describe something else entirely
Setup & Integrations
Technical difficulty
Essentially none. There is nothing to install, no extension to add, no configuration and no account to create. Using the tool means typing eight characters in front of a URL in the address bar, and the home page offers a form for anyone who prefers clicking. The only real prerequisite is access to one of the five supported AI assistants, and being signed in to it beforehand, since Claude in particular will otherwise divert to its login page.
Deployment
Integrations
Supported languages
Behind skim.page
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What data does skim.page actually see?
Does skim.page summarize the article itself?
Does it get past paywalls?
What can stop the handoff from working?
How much does it cost?
Which summary styles are available?
Which AI assistants does it support?
Is there an API, a mobile app or a browser extension?
Do I need an account?
How do I report a problem or ask for a feature?
Should you pick skim.page?
skim.page does one small thing and does it neatly. By turning a URL prefix into a ready-made summarization prompt, it removes the tedious step of copying a link, opening an assistant and typing out what you want done with it. The four angles are genuinely well-judged rather than decorative: the Research and Investor templates ask the questions a careful reader would actually ask, and the ELI5 template is a real aid for unfamiliar subjects. That the whole thing runs client-side, keeps no accounts and stores nothing is a sensible design for what it does.
Its value, though, is convenience rather than capability. skim.page produces no summary of its own; it is a well-shaped funnel into an assistant you already use, and its output is only as good as that assistant's ability to fetch and read the article. When the assistant cannot open the link, the tool has nothing to fall back on. The site is refreshingly honest about this, and about paywalls, which it explicitly declines to circumvent.
The reservations are about governance rather than function. No legal entity, no terms of service, no postal address, no contact email and no mention of the GDPR appear anywhere, and the only support channel is a public issue tracker. The repository carries no licence, sits at version 0.1.0 and has not been touched since May 2026, while the assistant handoff URLs it depends on can change at any time. Note also that the domain previously hosted an unrelated product, so older references to skim.page may not describe this tool.
As a free personal utility for everyday reading, it is a pleasant and well-made shortcut. For anything with procurement, compliance or continuity requirements, the missing paperwork is the deciding factor.
- Choosing a selection results in a full page refresh.
- Opens in a new window.