rtrvr v24: export Smart DOM Trees as Markdown

One cURL command. That's all it takes.

bash

curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: rtrvr_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "act",
    "params": {
      "user_input": "go to ChatGPT.com and extract citations for top restaurants"
    }
  }'

Your browser just became an API endpoint. No SDK. No complex setup. Just HTTP.

v24 is here, and it changes everything about web automation.

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

Your Chrome Extension is no longer just an extension—it's a full API server.

Remember the pain of browser automation? Puppeteer scripts breaking on every site update. Selenium grids consuming resources. Playwright fighting bot detection. All while your actual browser—logged into every service you need—sits there unused.

We flipped the model. Instead of simulating a browser in the cloud, we made YOUR browser programmable.

Before v24:

Install extension → Open popup → Type command → Wait for results
Great for ad-hoc tasks, impossible to integrate

With v24:

Install extension → Generate API key → Call from anywhere
Your CI/CD pipeline can now control your browser
Your Slack bot can trigger browser automation
Your cron job can execute browser tasks

The Extension now exposes a secure API endpoint. You can trigger it from:

Terminal via cURL
Python scripts
Node.js applications
n8n/Zapier/Make workflows
Literally any system that can make HTTP requests

bash

# Real example: Extract data and get structured JSON back
curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: rtrvr_j9DQpNvt94rJLOrmW0hu82pgZ4jDKz68tU99uEK_798" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "act",
    "params": {
      "user_input": "go to news.ycombinator.com and extract top 5 stories with scores"
    }
  }'

Your authenticated sessions. Your cookies. Your logged-in state. All accessible programmatically, securely, without credential sharing.

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

FireCrawl is dead. Markdown scraping is obsolete. Here's why.

Traditional scraping tools convert web pages to Markdown because that's what LLMs could historically understand. But this conversion is destructive:

What Markdown loses:

Hierarchical structure (nested elements flatten)
Interactive elements (buttons become text)
Layout relationships (grids become lists)
Semantic meaning (everything becomes headers and paragraphs)
JavaScript-rendered content (often missed entirely)

Our Smart DOM Trees preserve everything:

Full HTML structure with intelligent filtering
Interactive element states and properties
Spatial relationships and layout context
Semantic HTML5 tags and ARIA labels
Dynamic content loaded by JavaScript

bash

# Get Smart DOM Tree instead of broken Markdown
curl -X POST https://api.rtrvr.ai/scrape \
  -H "Authorization: Bearer rtrvr_YOUR_CLOUD_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "urls": ["https://complex-spa-website.com"]
  }'

The result? Our agents understand web pages the way browsers do—as structured documents, not walls of text.

Performance speaks for itself:

81.4% success rate on Halluminate Web Bench (vs 40-60% for screenshot-based agents)
88.24% success on data extraction tasks
0.9 minutes average completion time

This isn't incremental improvement. It's a fundamental rethinking of how AI agents interact with the web.

3. Enterprise-Ready Workflow Automation

v24 isn't just about individual features—it's about production-ready automation at scale.

File Upload Capability Agents can now upload files to any website. This unlocks:

End-to-end job applications (resume + cover letter)
Document submission workflows
Bulk data imports
Media uploads for content management

Shortcuts System Convert any successful action into a reusable shortcut:

Execute a complex workflow once
Save it as /apply-to-job or /competitor-check
Rerun instantly with one command
Share shortcuts across your team

Free MCP Tools We've exposed core browser control tools for developers using their own LLMs:

getPageData - Extract page content and structure
takePageAction - Control browser elements

Connect these to Claude, ChatGPT, or your own models. Use your existing AI subscriptions to power browser automation.

The Technical Revolution Under the Hood

Revamped Tree Construction

We rebuilt our DOM tree construction from the ground up:

50% less memory usage through intelligent pruning
Better coverage of complex SPAs and dynamic content
Faster processing with parallel tree traversal
Smarter element selection using semantic understanding

Getting Started with v24

For Extension Users

Update to v24 from the Chrome Web Store
Generate API Key in extension settings
Start calling your browser from anywhere

For Cloud API Users

No migration needed—just start using the new endpoints:

/agent for complex multi-step automation
/scrape for Smart DOM Tree extraction

The Bigger Picture: Infrastructure for the Agentic Web

v24 isn't just an update—it's infrastructure for a new computing paradigm.

We're moving from:

Apps that we click through → Agents that act for us
APIs that return data → APIs that perform actions
Cloud-first architecture → Browser-native intelligence

Your browser is the most authenticated, capable, and personal computer you own. v24 makes it programmable.

Start Building Today

v24 is live. Your browser is waiting to be programmed.

Quick Start:

bash

# Install extension → Generate API key → Run this:
curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: YOUR_KEY" \
  -d '{"tool": "act", "params": {"user_input": "your task here"}}'

Resources:

The future of web automation isn't about better scrapers or smarter bots. It's about making the browser itself intelligent and programmable.

Welcome to v24. Welcome to the programmable web.

Questions? Reach out on Discord or support@rtrvr.ai

One cURL command. That's all it takes.

bash

curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: rtrvr_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "act",
    "params": {
      "user_input": "go to ChatGPT.com and extract citations for top restaurants"
    }
  }'

Your browser just became an API endpoint. No SDK. No complex setup. Just HTTP.

v24 is here, and it changes everything about web automation.

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

Your Chrome Extension is no longer just an extension—it's a full API server.

We flipped the model. Instead of simulating a browser in the cloud, we made YOUR browser programmable.

Before v24:

Install extension → Open popup → Type command → Wait for results
Great for ad-hoc tasks, impossible to integrate

With v24:

Install extension → Generate API key → Call from anywhere
Your CI/CD pipeline can now control your browser
Your Slack bot can trigger browser automation
Your cron job can execute browser tasks

The Extension now exposes a secure API endpoint. You can trigger it from:

Terminal via cURL
Python scripts
Node.js applications
n8n/Zapier/Make workflows
Literally any system that can make HTTP requests

bash

# Real example: Extract data and get structured JSON back
curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: rtrvr_j9DQpNvt94rJLOrmW0hu82pgZ4jDKz68tU99uEK_798" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "act",
    "params": {
      "user_input": "go to news.ycombinator.com and extract top 5 stories with scores"
    }
  }'

Your authenticated sessions. Your cookies. Your logged-in state. All accessible programmatically, securely, without credential sharing.

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

FireCrawl is dead. Markdown scraping is obsolete. Here's why.

Traditional scraping tools convert web pages to Markdown because that's what LLMs could historically understand. But this conversion is destructive:

What Markdown loses:

Hierarchical structure (nested elements flatten)
Interactive elements (buttons become text)
Layout relationships (grids become lists)
Semantic meaning (everything becomes headers and paragraphs)
JavaScript-rendered content (often missed entirely)

Our Smart DOM Trees preserve everything:

Full HTML structure with intelligent filtering
Interactive element states and properties
Spatial relationships and layout context
Semantic HTML5 tags and ARIA labels
Dynamic content loaded by JavaScript

bash

# Get Smart DOM Tree instead of broken Markdown
curl -X POST https://api.rtrvr.ai/scrape \
  -H "Authorization: Bearer rtrvr_YOUR_CLOUD_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "urls": ["https://complex-spa-website.com"]
  }'

The result? Our agents understand web pages the way browsers do—as structured documents, not walls of text.

Performance speaks for itself:

81.4% success rate on Halluminate Web Bench (vs 40-60% for screenshot-based agents)
88.24% success on data extraction tasks
0.9 minutes average completion time

This isn't incremental improvement. It's a fundamental rethinking of how AI agents interact with the web.

3. Enterprise-Ready Workflow Automation

v24 isn't just about individual features—it's about production-ready automation at scale.

File Upload Capability Agents can now upload files to any website. This unlocks:

End-to-end job applications (resume + cover letter)
Document submission workflows
Bulk data imports
Media uploads for content management

Shortcuts System Convert any successful action into a reusable shortcut:

Execute a complex workflow once
Save it as /apply-to-job or /competitor-check
Rerun instantly with one command
Share shortcuts across your team

Free MCP Tools We've exposed core browser control tools for developers using their own LLMs:

getPageData - Extract page content and structure
takePageAction - Control browser elements

Connect these to Claude, ChatGPT, or your own models. Use your existing AI subscriptions to power browser automation.

The Technical Revolution Under the Hood

Revamped Tree Construction

We rebuilt our DOM tree construction from the ground up:

50% less memory usage through intelligent pruning
Better coverage of complex SPAs and dynamic content
Faster processing with parallel tree traversal
Smarter element selection using semantic understanding

Getting Started with v24

For Extension Users

Update to v24 from the Chrome Web Store
Generate API Key in extension settings
Start calling your browser from anywhere

For Cloud API Users

No migration needed—just start using the new endpoints:

/agent for complex multi-step automation
/scrape for Smart DOM Tree extraction

The Bigger Picture: Infrastructure for the Agentic Web

v24 isn't just an update—it's infrastructure for a new computing paradigm.

We're moving from:

Apps that we click through → Agents that act for us
APIs that return data → APIs that perform actions
Cloud-first architecture → Browser-native intelligence

Your browser is the most authenticated, capable, and personal computer you own. v24 makes it programmable.

Start Building Today

v24 is live. Your browser is waiting to be programmed.

Quick Start:

bash

# Install extension → Generate API key → Run this:
curl -X POST https://mcp.rtrvr.ai \
  -H "X-API-Key: YOUR_KEY" \
  -d '{"tool": "act", "params": {"user_input": "your task here"}}'

Resources:

The future of web automation isn't about better scrapers or smarter bots. It's about making the browser itself intelligent and programmable.

Welcome to v24. Welcome to the programmable web.

Questions? Reach out on Discord or support@rtrvr.ai

rtrvr v24: export Smart DOM Trees as Markdown

Dumping Markdown for Smart DOM Trees

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

3. Enterprise-Ready Workflow Automation

The Technical Revolution Under the Hood

Revamped Tree Construction

Getting Started with v24

For Extension Users

For Cloud API Users

The Bigger Picture: Infrastructure for the Agentic Web

Start Building Today

Roadmap: The Self-Improving Action Layer

AI Subroutines

Introducing agent analytics for Rover

Run the idea on a real site.

rtrvr v24: export Smart DOM Trees as Markdown

Dumping Markdown for Smart DOM Trees

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

3. Enterprise-Ready Workflow Automation

The Technical Revolution Under the Hood

Revamped Tree Construction

Getting Started with v24

For Extension Users

For Cloud API Users

The Bigger Picture: Infrastructure for the Agentic Web

Start Building Today

Roadmap: The Self-Improving Action Layer

AI Subroutines

Introducing agent analytics for Rover

Run the idea on a real site.

Dumping Markdown for Smart DOM Trees

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

3. Enterprise-Ready Workflow Automation

The Technical Revolution Under the Hood

Revamped Tree Construction

Getting Started with v24

For Extension Users

For Cloud API Users

The Bigger Picture: Infrastructure for the Agentic Web

Start Building Today

Send it their way.

Roadmap: The Self-Improving Action Layer

AI Subroutines

Introducing agent analytics for Rover

Run the idea on a real site.

Dumping Markdown for Smart DOM Trees

The Three Breakthroughs of v24

1. Remote Control Your Browser via API

2. Cloud API: Smart DOM Trees Kill Markdown Scraping

3. Enterprise-Ready Workflow Automation

The Technical Revolution Under the Hood

Revamped Tree Construction

Getting Started with v24

For Extension Users

For Cloud API Users

The Bigger Picture: Infrastructure for the Agentic Web

Start Building Today

Send it their way.

Roadmap: The Self-Improving Action Layer

AI Subroutines

Introducing agent analytics for Rover

Run the idea on a real site.