# Skim

> Skim is the x402-native Clean Reader API for AI agents. Send a URL, get back agent-ready Markdown plus structured metadata (title, byline, published date, language, excerpt) in about a second. Agents pay per call in USDC over HTTP 402 (the x402 protocol) on Base — no signup, no API keys, no subscription. Your wallet is your identity.

## Endpoints and prices

All endpoints share one payment flow: an unpaid request returns HTTP 402 with x402 payment instructions; an x402-capable client signs and retries. USDC on Base mainnet. Failed requests (4xx/5xx) do not settle — you are never charged for an error.

- GET https://skim402.com/api/v2/read?url=<encoded URL> — $0.002. Clean Markdown + metadata from any page. Options: &chunked=true (free — adds RAG-ready chunks with heading breadcrumbs and approximate token counts), &stripLinks=true, &stripImages=true (leaner Markdown).
- GET https://skim402.com/api/v2/read/js?url=... — $0.005. Same output, with full JavaScript rendering for SPAs and script-heavy pages.
- POST https://skim402.com/api/v2/read/batch — $0.020. Body { "urls": [...] }, up to 10 URLs read in parallel; per-URL results, partial successes still return.
- POST https://skim402.com/api/v2/extract — $0.015. Body { "url", "schema", "instructions"? }. Typed JSON matching your own JSON Schema; values come only from the page, never invented. Fields the page does not state come back null.
- GET https://skim402.com/api/v2/extract/{preset}?url=... — $0.015. Six ready-made schemas: article, product, job, review, event, table. Optional &instructions= hint.
- GET https://skim402.com/api/v2/tables?url=... — $0.003. Every genuine HTML table on the page parsed mechanically (no LLM): header-keyed records plus GitHub-flavored Markdown per table. Zero tables returns 422 with no charge.
- POST https://skim402.com/api/v2/dataset — $0.35 per build. Body { "prompt", "fields": [up to 12 column definitions], "urls": [1-10 seed URLs], "rowLimit"? }. You define the columns — no presets. Skim reads the seeds, follows relevant links, and returns structured rows, each with a sourceUrl citation. Returns a 202 job ticket; polling GET /api/v2/dataset/{id} is FREE. Pagination: when a build stops early the poll response has "hasMore": true and a "continuation" block — POST { "continueFrom": "<datasetId>" } starts a new $0.35 build that resumes exactly where the last one stopped, never repeats rows, and chains until hasMore is false. Nothing-left-to-continue rejects before payment settles.
- POST https://skim402.com/api/v2/watch — $0.01. Body { "urls": [1-20 URLs], "note"? }. Registers a custom watch list and returns a private watch id (the id is the access token — keep it secret). Then GET https://skim402.com/api/v2/watch/diff?id=<watchId> — $0.005 per poll — re-reads every URL and answers "what changed since my last check": per-URL status (changed / unchanged / first_check / error), added and removed line samples from the clean text (never raw HTML noise), a change ratio, and a numericOnly flag that separates price/counter ticks from real content changes. First poll baselines each page. Polls within 60 seconds of the previous one return the cached result ("fresh": true). GET /api/v2/watch/status?id= is FREE (registration info + poll count, no diff).
- GET https://skim402.com/api/v2/feeds/x402/latest — $0.005 per poll. The x402 ecosystem feed: new paid services and price changes across public x402 registries.
- GET https://skim402.com/api/v2/signal/{name}/latest — $0.005 per poll. Fifteen vertical intelligence feeds (the Skim Signal series; the sixteenth signal is the x402 ecosystem feed above), each returning structured newest-first items with links and extracted entities: ai-news, sec-filings (?forms=), crypto-news, macro, security, regulations, courts, recalls, deals (?categories=), launches, trending, research (?fields=), energy, entertainment, campaign-finance (?forms=&committees=). Catalog: https://skim402.com/signals
- Legacy /api/v1/* paths still work at the same prices for v1-generation x402 clients (POST bodies instead of GET query params).

## Key facts

- Output is clean Markdown, roughly 4x smaller than the raw HTML — no ads, nav, or boilerplate. Cheaper LLM calls on any model.
- In production benchmarks Skim returned pages consistently faster than Firecrawl, with every call backed by an on-chain USDC receipt on Base.
- Works with any x402 client (x402-fetch for JS, the official x402 Python SDK). Framework connectors on PyPI: langchain-skim, llama-index-readers-skim, crewai-skim, skim-haystack. MCP server: skim-mcp (npm) and io.github.JessieJanie/skim402 (official MCP Registry).
- No wallet? The official Apify Actor (apify.com/jessiejanie/skim-clean-reader) offers the same reads with normal Apify billing at $4.00 per 1,000 pages, platform usage included.
- Free browser demo: https://freeskims.skim402.com — 10 free skims a day, no signup.
- Skim reads public pages only: no login walls, no anti-bot racing, no credentials held — a least-privilege reader by design.

## Docs

- [Agent integration page](https://skim402.com/agents): the complete endpoint/price catalog plus copy-paste tool definitions for OpenAI, Anthropic, Gemini, Vercel AI SDK, LangChain, and MCP
- [Developer docs](https://skim402.com/docs): request/response shapes, x402 payment flow, code samples in curl, JavaScript, and Python
- [OpenAPI spec](https://skim402.com/openapi.json): machine-readable API contract for the core read/extract endpoints
- [Signals catalog](https://skim402.com/signals): all 16 signals (15 vertical feeds + the x402 ecosystem feed), with sources
- [Pricing](https://skim402.com/pricing): per-call prices, no subscription
- [Performance benchmarks](https://skim402.com/performance): methodology and live speed comparison
- [Comparison vs alternatives](https://skim402.com/compare): Skim vs Firecrawl, Jina, and other readers
- [Skim vs Firecrawl](https://skim402.com/compare/firecrawl), [Skim vs Jina](https://skim402.com/compare/jina), [Skim vs Exa](https://skim402.com/compare/exa)
- [Cost calculator](https://skim402.com/savings): DIY scraping vs Skim, interactive
- [Q&A](https://skim402.com/faq): common questions from developers

## About

- x402 protocol: https://x402.org
- Skim is listed on x402scan, the x402 Bazaar, the x402.org ecosystem directory, the official MCP Registry, and the Apify Store.
- Contact: hello@skim402.com or https://skim402.com/contact