Context.dev Launches 'Scrape Anything': Unified Web and Document Markdown Extraction API

On September 23, 2026, developer tooling provider Context.dev officially unveiled 'Scrape Anything', a consolidated data extraction API designed to merge web scraping and multi-format document conversion into a single, unified REST endpoint while handling anti-bot protections and JavaScript rendering entirely on the server side.

Image source: @mynameisyahia on X / Context.dev

As large language models (LLMs) and retrieval-augmented generation (RAG) pipelines have become foundational across software engineering, maintaining reliable data ingestion pipelines has emerged as a persistent operational headache. Engineering teams typically maintain complex infrastructure combinations: headless browser clusters and residential proxy rotations for dynamic web pages, juxtaposed with separate parsing libraries to extract content from PDFs, spreadsheets, and slide presentations. Scrape Anything aims to eliminate this fractured maintenance burden by providing an all-in-one ingestion gateway where any URL yields clean, standardized output through a single API call.

Consolidating Web Scraping and 40+ Document Formats into One Endpoint

The defining technical advantage of Scrape Anything is its format-agnostic ingestion engine. Rather than requiring distinct parsers for different MIME types, the endpoint transparently accepts web URLs alongside more than 40 enterprise and document file formats.

Developers can submit URLs pointing directly to live websites or unstructured files, including PDF, DOCX, DOC, XLSX, XLS, PPTX, PPT, CSV, XML, JSON, and SVG. The API parses the underlying structure and returns clean, LLM-ready GitHub Flavored Markdown (GFM) as its primary output:

• Preserving Spreadsheet Structure: Multi-sheet Excel workbooks (XLSX, XLS) are parsed into distinct GFM tables per sheet, with each sheet name formatted under an ## heading. Flat CSV files are similarly converted into clean, aligned markdown tables.
• Hierarchical Slide Extractions: PowerPoint presentations (PPTX, PPT) are systematically extracted into sequential slide units (## Slide N: Title), preserving body text, structured tables, and underlying speaker notes.
• Automated Content Cleaning: Passing the parameter useMainContentOnly: true prompts the parser to identify and strip boilerplate elements—such as top navigation bars, footers, and sidebars—returning solely high-signal body text to optimize LLM token consumption.
• Multiple Output Formats: In addition to standardized GFM, the API supports raw rendered HTML, structured JSON responses, full-page screenshots, and image asset manifests.

Server-Side Anti-Bot Bypass and Headless Infrastructure

Web scraping at scale frequently breaks down at the network edge due to aggressive anti-bot defenses like Cloudflare and hCaptcha, coupled with client-side JavaScript rendering barriers like iframes and Shadow DOMs.

Context.dev offloads this operational friction entirely to its backend infrastructure. Upon receiving a request, the service manages intelligent proxy escalation and rotation, spins up headless browsers for complete client-side execution, unwraps nested iframes and shadow DOM trees, and executes internal backoff-retry logic. Developers receive clean parsed output without configuring local Chromium instances or managing third-party proxy pools.

Initial community reactions also included individual anecdotal feedback. On X, developer Joey (@jsmillerdev) shared that in an informal test across roughly 130 sites previously blocked by Cloudflare defenses, Scrape Anything successfully processed 95% of them, noting that he was able to integrate and call the API within a minute of onboarding.

API Architecture and Work-Email Free Tier Policy

Scrape Anything exposes a straightforward, RESTful interface designed for frictionless integration into existing backend services and agent frameworks:

• GET /web/scrape/markdown: Takes any URL or supported document link and returns cleaned GitHub Flavored Markdown.
• GET /web/scrape/html: Fetches fully rendered raw HTML following complete client-side script execution.
• GET /web/scrape/images: Returns a structured manifest of image assets located on the target page.
• GET /web/scrape/sitemap: Discovers and enumerates all URLs indexed in a target domain's sitemap.
• POST /web/crawl: Initiates domain-wide crawling capped by configurable maxPages and maxDepth parameters.

To support developer adoption, Context.dev provides a recurring free tier granting 1,000 requests per month (1k reqs/mo) to accounts registered with a corporate or work email address.

Engineering teams should note several key operational boundaries. The 1,000 monthly request quota is strictly reserved for business domains and is not granted to consumer email addresses like Gmail. Furthermore, domain-scale ingestion requires invoking the dedicated /web/crawl endpoint rather than repetitive single-scrape calls. Finally, targets protected behind strict session authentication or complex multi-step interactive workflows still require custom header passing and specialized pre-processing configurations.

Sources

• Context.dev Web Scraping API Reference
• Context.dev Scrape Websites to Markdown Guide
• Yahia Bakour Announcement on X (@mynameisyahia)