ContextCut — slice web pages into clean AI context
Crop web pages into clean Markdown snippets and feed them to AI within a token budget.
- Problem
- When feeding web material to AI coding agents, full-page scraping drags in navigation, ads, and noise that waste precious context window. One developer built Markdownee to crawl sites into Markdown and another made a TypeScript port, showing this need keeps appearing.
- Solution
- A browser extension plus CLI: select a region on a page, auto-extract the main content into clean Markdown, trim and dedupe to fit a token budget, and copy or send it into Claude / Codex context files with one click.
- Users
- Developers researching with AI coding agents, technical writers, PMs who feed material to models.
- MVP
- Browser region selection and main-content extraction; HTML to clean Markdown; token counting and budget hints; copy or save as local .md; support common docs sites and blogs.
- Revenue
- Free extension plus a $5/mo tier for batch capture, shared team snippet libraries, and direct agent integration.
- Why now
- AI agents burn context like money, so precise feeding is increasingly valued; existing tools either crawl whole sites or only convert, missing on-demand slicing.