JUNCTION 2026Community
All challenges

Junction 2026 · Tech challenges

Best Use of Context.dev: Fully Structured Web

by Context.dev

Plug the Context.dev API into whatever you're building and give it the live web: pages, brands, companies, people and news as clean structured data in one API call.

The Challenge

Whatever you're building, use the Context.dev API to extract exactly what you need from the web and nothing more.

Live web data lets your project check facts against current sources, react when a page or news story changes, brand itself from a customer's email, or enrich a company or person in one call instead of relying on stale data.

To count, your project must call at least one Context.dev endpoint in its core flow, not just as a side feature.

Minimum requirements

  • Use at least one Context.dev API endpoint in the submitted project.
  • The data from Context.dev must visibly drive something in the demo.

What to submit

Submit a GitHub repository and a demo of your project, plus a short note on which Context.dev endpoints you used and why.

Resources

What we'll bring

Every team gets 50,000 Context.dev API credits that don't expire, plus documentation, SDKs, an MCP server, live example projects and Context.dev engineers on site to help.

No credit card needed. There is no rate limit, with a maximum of 50 concurrent requests per organisation. No special hardware is needed; the live web is your dataset.

The Technology

Context.dev is a web data API. You send it a URL, a domain, a company or a search query, and it returns clean, structured data — markdown, JSON, screenshots, logos, colours, company facts and news — that your app or model can use right away. No scrapers, proxies or parsers to maintain.

Key capabilities:

  • Search and scrape: turn pages into markdown, HTML, screenshots or images, or crawl and map a whole site.
  • Structured extraction: extract JSON with your own schema, or get researched answers to a question.
  • Brand data: get logos, colours, fonts and a style guide from a domain or email.
  • Search and enrichment: search and enrich companies, people and news.
  • Monitors and webhooks: react when a page or topic changes.
  • Batch processing: process thousands of URLs.
  • File parsing: parse PDFs and documents.
  • Social media: scrape feeds such as TikTok and Instagram.

Getting started

  1. Create an account at context.dev/signup. Coding agents can use context.dev/auth.md.
  2. Redeem code JUNCTION26 at Redeem credits to receive 50,000 API credits per team, with no expiry.
  3. Copy a secret key from the API keys page.
  4. Install the SDK: npm install context.dev or pip install context.dev.
  5. Scrape your first page using the code in the Quickstart. You should see the Junction homepage printed as Markdown. Swap in any URL, or try brand, search, news or monitors. Start building!

Using a coding agent? Point it at the Agent quickstart, or connect the MCP server at https://mcp.context.dev/mcp (setup guide).

Example requests and responses:

Recommended setup

Use any language over HTTP. SDKs are available for TypeScript/Node, Python, Ruby and Go. If you can call an API, you can use Context.dev, whether you're building a web app, backend or something else that can make an HTTP request.

Bring any laptop with Node.js 20+ or Python 3.9+ (Ruby 3.2+ and Go 1.22+ are also supported), or just curl. Internet access is required.

Optional tools: the context-dev CLI, the agent skill, or the MCP server. To try requests without code, use the Playground.

Tips and common pitfalls

  • Check each output's success flag. A 200 response can still contain a failed output.
  • Set maxAgeMs: 0 when you need fresh data. Results are cached for up to 3 days by default, and cached results cost the same credits as fresh ones. Store results in your own app to avoid paying for repeat calls.
  • Crawls, batches and monitors are asynchronous. Poll the job or receive a webhook; use a service such as Cloudflare Tunnel for local webhooks.
  • Some sites block automated access. Teams are responsible for respecting the terms of the sites they use.

Documentation and examples

There is no starter repository; start with the Quickstart guide.

Mentors and support

Context.dev engineers will help with web data extraction, search, LLM apps and large-scale data pipelines:

  • Beer van der Drift — CTO, challenge lead and lead judge
  • Yahia Bakour — Founder & CEO
  • Thomas Oomens — Engineer

On-site availability:

  • Friday: until 23:00
  • Saturday: from 09:00 until Sunday 02:00
  • Sunday: from 07:00 until the end

Come visit the Context.dev booth. Main and technical contact: Beer van der Drift, CTO, beer@context.dev. Service status: status.context.dev.