chrisrobison

chrisrobison / textweb

Public

A text-grid web renderer for AI agents — see the web without screenshots

72
6
100% credibility
Found Feb 19, 2026 at 43 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

TextWeb renders web pages as compact, structured text grids that preserve spatial layout and annotate interactive elements, allowing AI agents to browse and interact without screenshots or vision models.

How It Works

1
🔍 Discover TextWeb

You hear about TextWeb, a smart way to let AI helpers explore websites using simple text views instead of pictures.

2
📥 Set it up quickly

You grab the tool and get it ready on your computer in moments, no hassle involved.

3
🌐 View your first page

You point it at any website and instantly see a neat text map of the page, with clickable spots marked clearly.

4
🤖 Link it to your AI

You connect it easily to your favorite AI assistant, like a simple add-on that unlocks web browsing.

5
🖱️ Watch AI interact

Your AI now reads the text layout, clicks links, types in forms, and scrolls around just like you would.

✅ AI browses freely

Your assistant handles web tasks smoothly and quickly, saving time and making everything feel effortless.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 43 to 72 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is textweb?

TextWeb is a JavaScript-based text web renderer that turns any website into a compact, structured text grid—think a terminal-style view of the full page layout, complete with interactive element references like [1] for links and inputs. Built on Playwright for real browser rendering, it lets AI agents "see" dynamic web UIs without massive screenshots or vision models, outputting just 2-5KB of parseable text. CLI tools render pages instantly (`textweb https://example.com`), while an HTTP API and Node.js library enable navigate, click, type, scroll actions via simple POSTs.

Why is it gaining traction?

Unlike accessibility trees that lose spatial layout or raw HTML dumps that overwhelm LLMs, TextWeb preserves pixel-accurate positioning in a grid agents can reason over natively—"click ref 9" just works. It's blazing fast and cheap, dodging vision model costs, with plug-and-play integrations for MCP clients (Claude Desktop, Cursor), LangChain, CrewAI, and OpenAI tools. Devs love the interactive REPL mode and JSON outputs for quick agent prototyping.

Who should use this?

AI agent builders wiring LLMs to browse sites, like researchers automating Hacker News scraping or job hunters filling forms via CrewAI tasks. Backend devs adding text webui browsing to Spring Boot apps without screenshots, or frontend teams testing text-grid views of SPAs. Perfect for LangChain zero-shot agents needing reliable web interaction.

Verdict

Grab it for agent prototypes—solid docs, MIT license, and npm-ready make setup trivial despite 33 stars and 1.0% credibility signaling early maturity. Polish tests and add more examples to hit escape velocity.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.