BotBrowser – MCP server, saves 90% of tokens for web-browsing agents

Name: BotBrowser – MCP server, saves 90% of tokens for web-browsing agents
Availability: InStock
Author: amplifyco

by amplifyco·Feb 27, 2026·1 point·1 comment

Visit Project View on HN

AI Analysis

●●●BangerSolve My ProblemWizardryDark Horse

Strips 90% of tokens from web pages for agents—no API key, no server, MIT open source.

Strengths

•Battle-tested extraction engines (Mozilla Readability, Trafilatura) borrowed from production readers, not custom regex hacks
•Zero-setup philosophy—npm/pip install with no API key, auth, or server requirement removes friction for local agent workflows
•Dual native SDKs in JS and Python mean no thin wrapper nonsense; each implementation optimized for its ecosystem

Weaknesses

•Hosted MCP tier ($29/mo) competes with Firecrawl and Jina.ai, which offer similar token savings with richer feature sets
•JS rendering limited to 300/day on Pro tier, which may constrain heavy browsing agents without upgrade to Enterprise

Post Description

Hi HN, I built BotBrowser because I was tired of watching AI agents burn through tokens or context on raw HTML — scripts, ads, nav bars, cookie banners — just to read an article or docs.

BotBrowser gives your agent clean Markdown from any web page. A typical 50K token page comes back as ~3K tokens with all the useful content preserved.

It works as an MCP server — add one line to your Claude Desktop or Cursor config and your agent can browse the web, search and extract multiple pages, render JavaScript-heavy sites, and even take screenshots.

We're live at https://thebotbrowser.com — free tier, no credit card required.

Happy to answer any questions.