Agentic Browser Chat

Tracking since Jun 4, 2026.

Changelog

• Jun 9, 2026
  description

  Agentic Browser Chat puts an AI assistant right next to the page you're reading. Ask questions about any web page, take notes, manage tasks, generate documents and images, and let the assistant act on the page for you — all without leaving the tab.
  
  You bring your own OpenRouter API key, so you choose the model (Claude, GPT, Gemini, Llama, and many more) and pay only for what you use. Everything you create is stored locally in your browser. There is no account, no developer server, and no analytics.
  
  ━━━━━━━━━━━━━━━━━━━━
  WHAT YOU CAN DO
  ━━━━━━━━━━━━━━━━━━━━
  
  CHAT WITH ANY PAGE
  • Open the panel on any site and ask about what's on screen
  • Right-click selected text to Explain, Summarize, Proofread, or ask a Quick Question
  • Use the content selector to point at any element and add just that piece to the chat
  • Searchable, pinnable chat history with per-conversation model choice
  
  AN ASSISTANT THAT TAKES ACTION
  • Search the web and fetch pages (with safeguards that stop the model from inventing URLs)
  • Fill forms — with a confirmation step before anything is submitted
  • Run precise calculations in a sandboxed, network-isolated worker
  • Generate downloadable DOCX, PPTX, XLSX, PDF, and CSV files
  • Generate images inline
  
  NOTES, TASKS & QUIZZES IN ONE PLACE
  • Notes with version history, a pop-out editor, and file attachments
  • Tasks with due dates and reminder notifications
  • Auto-generated quizzes (multiple choice and fill-in-the-blank) for studying
  
  ATTACH ANYTHING
  • PDF, DOCX, PPTX, XLSX, CSV, images, and more
  • Capture a screenshot, pull in another tab's content, or paste a spreadsheet
  • Attach an existing note or a summary of another chat as context
  
  REMEMBERS WHAT MATTERS
  • Persistent memory: ask it to remember facts and preferences across every chat
  • Custom instructions applied to all conversations
  • Reusable "skills" — save a procedure once and have the assistant follow it on demand
  
  ━━━━━━━━━━━━━━━━━━━━
  PRIVACY FIRST
  ━━━━━━━━━━━━━━━━━━━━
  
  • All your data (chats, notes, tasks, attachments, settings) stays on your device
  • The only external connection is to OpenRouter, using your own API key
  • No tracking, no telemetry, no developer-controlled server, no data resale
  
  ━━━━━━━━━━━━━━━━━━━━
  GETTING STARTED
  ━━━━━━━━━━━━━━━━━━━━
  
  1. Install and pin the extension
  2. Open the panel and paste your OpenRouter API key in Settings
  3. Pick a model and start chatting about any page
  
  Open source. Source code and full privacy policy are available on GitHub.

  Agentic Browser Chat puts an AI assistant right next to the page you're reading. Ask questions about any web page, take notes, manage tasks, generate documents and images, and let the assistant act on the page for you, all without leaving the tab.
  
  You bring your own OpenRouter API key, so you choose the model and pay only for what you use. Everything you create is stored locally in your browser. There is no account, no developer server, and no analytics.
  
  ━━━━━━━━━━━━━━━━━━━━
  WHAT YOU CAN DO
  ━━━━━━━━━━━━━━━━━━━━
  
  CHAT WITH ANY PAGE
  • Open the panel on any site and ask about what's on screen
  • Right-click selected text to explain, summarize, proofread, or ask a quick question
  • Use the content selector to point at any element and add just that piece to the chat
  • Searchable, pinnable chat history with a per-conversation model choice
  
  AN ASSISTANT THAT TAKES ACTION
  • Search the web and fetch pages, with safeguards that stop the model from inventing URLs
  • Fill forms, with a confirmation step before anything is submitted
  • Run precise calculations in a sandboxed, network-isolated worker
  • Generate downloadable documents, spreadsheets, presentations, and PDFs
  • Generate images inline
  
  NOTES, TASKS & QUIZZES IN ONE PLACE
  • Notes with version history, a pop-out editor, and file attachments
  • Tasks with due dates and reminder notifications
  • Auto-generated quizzes for studying, in multiple choice and fill-in-the-blank formats
  
  ATTACH ANYTHING
  • Add documents, spreadsheets, and images to give the assistant context
  • Capture a screenshot, pull in another tab's content, or paste a spreadsheet
  • Attach an existing note or a summary of another chat as context
  
  REMEMBERS WHAT MATTERS
  • Persistent memory: ask it to remember facts and preferences across every chat
  • Custom instructions applied to all conversations
  • Reusable skills: save a procedure once and have the assistant follow it on demand
  
  ━━━━━━━━━━━━━━━━━━━━
  PRIVACY FIRST
  ━━━━━━━━━━━━━━━━━━━━
  
  • All your data stays on your device, including chats, notes, tasks, attachments, and settings
  • The only external connection is to OpenRouter, using your own API key
  • No tracking, no telemetry, no developer-controlled server, no data resale
  
  ━━━━━━━━━━━━━━━━━━━━
  GETTING STARTED
  ━━━━━━━━━━━━━━━━━━━━
  
  1. Install and pin the extension
  2. Open the panel and paste your OpenRouter API key in Settings
  3. Pick a model and start chatting about any page
  
  Open source. Source code and full privacy policy are available on GitHub.

Permissions & access

Permissions

contextMenusactiveTabtabsscriptingclipboardWriteclipboardReadstoragealarmsnotificationsoffscreen

Host access

<all_urls>

Screenshots