Browser to Terminal

Bridge your browser to AI coding agents

Screenshots, elements, and errors — delivered to your terminal AI in one click.

Want to try first? Start 2-day free trial

{ }
Agent Logger
Online
Claude Code
terminal_v...
Connected to
terminal_vision
Claude Code MCP Setup
Native MCP
Errors0
AllErrorsWarningsNetwork
Search errors...
No errors detected
Works With
Claude Code
Cursor
Windsurf
Aider
Gemini CLI
VS Code
Screenshot Capture

Capture and send to your AI instantly

Take viewport, full-page, region, or element screenshots. Preview and send to your terminal AI in one click.

localhost:3000
Features

Everything your AI needs to see the browser

Complete browser context for your AI coding assistant.

Element Selection

Click any element to get selector, dimensions, and styles.

Screenshot Capture

Four capture modes: viewport, full-page, region, or element.

Error Monitoring

Real-time console errors with smart noise filtering.

Browser Automation

Let your AI click, type, and scroll through conversation.

Accessibility Auditing

Inspect ARIA roles and run comprehensive a11y audits.

Auto-Push Events

Stream browser events directly to your AI as they happen.

Element Inspector

Select any element, get full context

Click any element on the page to capture its selector, dimensions, styles, and accessibility info — then send it to your AI.

localhost:3000
Features

Everything your AI needs

Complete browser context for your coding assistant.

<div.flex.flex-col.space-y-1.5.p-6>
366 × 208

Element Selection

Click any element to get selector, dimensions, and styles.

Screenshot Capture

Four capture modes: viewport, full-page, region, or element.

Error Monitoring

Real-time console errors with smart noise filtering.

Browser Automation

Let your AI click, type, and scroll through conversation.

Accessibility Auditing

Inspect ARIA roles and run comprehensive a11y audits.

Auto-Push Events

Stream browser events directly to your AI as they happen.

Error Monitoring

Catch every error, help your AI debug

Real-time console errors, network failures, and exceptions — automatically captured and ready to send to your AI for instant debugging.

app.example.com/dashboard

Dashboard

Welcome back, Developer

Settings
New Project
1,234
Total Users
Error
Failed to load
89%
Uptime
Timeout
Slow response

Failed to load documents

Error loading documents: Cannot read properties of undefined (reading 'map')

Uncaught TypeError: Cannot read properties of undefined (reading 'map')
Smart Error Context

Errors with exact source location

Your AI gets pinpoint accuracy — file path, line number, and stack trace. No more searching the entire codebase. Just fix and move on.

Without Smart Context
Traditional error logging
TypeError: Cannot read properties of undefined
No source information available
Search entire codebase for error
Read multiple files to find source
Multiple back-and-forth iterations
Wastes tokens on searching
With Agent Logger
Smart error context
TypeError: Cannot read properties of undefined
at fetchUserData (src/api/users.ts:47:12)
at Dashboard (src/components/Dashboard.tsx:23:5)
Exact file and line number provided
Full stack trace with context
Jump directly to the problem
Optimized token usage
#0
Error Details
Indexed for quick AI reference
CRITICALjust now
Error Message
TypeError: Cannot read properties of undefined (reading 'map')
Source Location
src/api/users.ts:47:12
Function
fetchUserData()
Stack Trace
1
at fetchUserData (src/api/users.ts:47:12)
2
at Dashboard (src/components/Dashboard.tsx:23:5)
3
at renderWithHooks (react-dom.js:14985:18)
4
at mountIndeterminateComponent ...
Your AI knows exactly where to look
With indexed errors and precise source locations, your coding agent can jump straight to src/api/users.ts:47 — no searching, no guessing, no wasted tokens.
10x
Faster error resolution
90%
Less searching required
#0
Indexed for quick reference
MCP Native

Your AI controls the browser. Directly.

Built on the Model Context Protocol. Your terminal AI doesn't just receive data — it can request screenshots, click elements, and automate the browser through natural conversation.

Terminal AI

Claude Code, Cursor, Windsurf, Aider — any MCP-compatible agent

screenshot
click
get_errors

MCP Protocol

Bidirectional communication — request & receive in real-time

Browser

Agent Logger extension with full page access

image_data
element_info
errors[]

What your AI can do from the terminal

No manual screenshots. No copy-paste. Just ask.

Capture Screenshots

Request viewport, full-page, or element screenshots instantly

browser_screenshot

Click & Type

Interact with page elements through natural commands

browser_click

Get Errors

Fetch console errors with full stack traces on demand

browser_get_errors

Automate Flows

Scroll, hover, and navigate — all through conversation

browser_scroll
claude-code
U

The button on the homepage isn't working. Can you check what's wrong?

Let me check the browser for errors...

→ browser_get_errorsfetching...

Found it! There's a TypeError at src/components/Button.tsx:23. The onClick handler is undefined. Let me fix that...

Zero context switching. Your AI debugged and located the issue without you leaving the terminal.
Features

Everything your AI needs to see the browser

Complete browser context for your AI coding assistant.

Element Selection

Click any element on the page. Get selector, dimensions, computed styles, and accessibility info — all sent to your AI in one click.

Selected element
selector: "#submit-btn"
tag: button
text: "Submit"

Screenshot Capture

Four modes: viewport, full-page, region, or element. Annotate with arrows, text, and highlights.

ViewportFullRegion

Error Monitoring

Real-time console errors, exceptions, and network failures with full stack traces.

Live monitoring

Browser Automation

Let your AI take control. Click buttons, fill forms, hover elements, and scroll pages — all through natural conversation.

clicktypescrollhover
Airplane Ready

Works Offline

No internet? No problem. Capture screenshots, select elements, and keep working. Auto-syncs when you're back online.

7-day offlineAuto-syncZero downtime

Accessibility Auditing

Inspect ARIA roles, labels, and states. Run comprehensive a11y audits with actionable reports.

WCAG 2.1 compliant

Auto-Push Events

Enable auto-push mode to stream browser events, errors, and selections directly to your AI as they happen.

Real-time streaming

Page Context

Automatically include page URL, title, and metadata with every interaction for complete context.

localhost:3000/dashboard
Offline Mode

Code anywhere. Even at 35,000 feet.

Your session stays active for 7 days without internet. Work from planes, trains, or that coffee shop with terrible WiFi.

On a flight
Code at cruising altitude
On a train
Work through tunnels
Spotty WiFi
Never lose progress
Without Offline Mode
Connection required
Connection Lost
Session expired. Please reconnect to continue.
Forced logout when offline
Can't capture screenshots
Work progress interrupted
Productivity lost
With Agent Logger
Offline mode active
Offline Mode
6 days left
All features available. Auto-sync when back online.
Session stays active for 7 days
Full feature access offline
Cryptographic session validation
Zero productivity loss

How Offline Mode Works

1. Login Online
Sign in once to receive your secure offline token
2. Work Offline
Use all features for up to 7 days without internet
3. Auto-Sync
Reconnect and your token refreshes automatically
Cryptographically secured
Signed tokens verified locally — your session is protected even offline
7
Days offline access
100%
Feature availability
0
Productivity loss
Setup

Three steps. Zero configuration.

Get up and running in under 2 minutes.

1

Install

Add the Chrome extension from the Web Store.

# Install from Chrome Web Store
# or download from the official website
2

Setup

Run the setup wizard to configure the MCP server and native messaging.

node setup.js
3

Connect

Open the side panel, enter the pairing code, and start sending context to your AI.

claude mcp add agent-logger \
node ./mcp-server/server.js
Ready! Open the side panel and start sending context.
Terminal Support

Works with your terminal

Seamless text injection for tmux and iTerm2. Clipboard fallback for everyone else.

Full Support

tmux

Method: send-keys

Text appears instantly in your active pane

Full Support

iTerm2

Method: AppleScript

Native macOS integration with instant injection

Clipboard

Terminal.app

Method: Clipboard

Copies to clipboard - press Cmd+V to paste

Clipboard

VS Code

Method: Clipboard

Copies to clipboard - paste into integrated terminal

💡

How terminal detection works

Agent Logger automatically detects your terminal type via environment variables ($TMUX, $ITERM_SESSION_ID, $TERM_PROGRAM). The extension UI adapts to show "Send" or "Copy" based on injection capability.

Limited Time Offer

Stop debugging. Start building.

Every minute hunting errors is a minute stolen from creating.

100+ developers joined
Launch price ending soon
Early Adopter Special

One payment. Yours forever.

No subscriptions. No premium tiers. Just unlimited access to everything — now and forever.

Lifetime access forever
All future updates free
Full feature access
Works offline for 7 days
Priority support
Early access to new features
Secure checkoutInstant accessNo recurring fees
SAVE $83/YEAR
$9/month
$25
one-time payment
yours forever

One-time payment. Yours forever.
No subscriptions, no recurring fees.

Price increases after launch period
Lock in $25 — this price won't return

You became a developer to create, not to hunt through stack traces.Your next feature is waiting.

2-Day Free Trial

Not ready to commit?

Try Agent Logger free for 2 days. Full features, no limitations.
Then just $25 — forever.

2
DAYS

Payment details required. Cancel anytime.

Full feature accessWorks offlineAutomatic upgrade
Get Started

Ready to bridge your browser to AI assistants?

Install the extension and start sending context to Claude Code in seconds.

or start with a 2-day free trial

Chrome, Chromium, Brave, or EdgeNode.js 18+macOS, Linux, or Windows