TuClaw - AI Agent to Automate Browser Tasks

Tracking since May 7, 2026.

Changelog

• May 30, 2026
  name

  TuClaw - Your Personal AI Agent

  TuClaw - AI Agent to Automate Browser Tasks

• May 23, 2026
  short_description

  Books appointments, sends emails, scans listings, fills forms. A browser agent that lives in your side panel.

  AI agent that automates web tasks in your browser. Describe a task. AI does the clicking and typing.

• May 23, 2026
  name

  TuClaw - Personal AI Assistant

  TuClaw - Your Personal AI Agent

• May 17, 2026
  description

  AI Web Agent is a chat-driven browser automation tool that lives in Chrome's side panel. Describe a task in plain English. It clicks, types, navigates, and reports back.
  
  Everything happens in your real browser session, so accounts you're already signed into work without copying credentials anywhere new. The agent reads structured page content alongside a screenshot, so it understands what is actually on the page and adapts when sites change their layout.
  
  Bring your own API key. The extension supports Claude (Anthropic), ChatGPT (OpenAI), and Gemini (Google). You pay your provider directly, no markup, no middleman. Switch models any time from the dropdown.
  
  What you can do with it:
  
  Chat with your browser. Tell it what you need it does the searching, clicking, and reading.
  
  Reference specific tabs with @. Type @ in the chat input and pick the tab you want the agent to work on. Useful when you have research spread across ten tabs and want to be precise.
  
  Safe mode. Before any irreversible action like submitting a form, posting something public, or deleting data, the agent stops and asks for permission. You see what it is about to do and approve or reject.
  
  Read pages, not pixels. The agent pulls clean text from articles, emails, and docs instead of trying to interpret a screenshot. It actually reads the content.
  
  Idle awareness. Opt in, and scheduled tasks defer until you stop using the browser, so the agent does not hijack your screen mid task.
  
  Max features:
  
  Scheduled tasks. Run something every hour, every Friday at 9am, or once in 30 minutes. The chat agent decides whether your scheduled prompt is a reminder ("nudge me to stretch") or a real browse task ("check my Gmail and summarize what is new"), so reminders surface as notifications.
  
  Persistent memory. Each session is compressed into semantic notes organized in a memory palace structure, with topic wings and rooms. Relevant notes load automatically when you start a related task. The agent gets sharper at recurring work over time without you copy pasting context.
  
  Scheduled run history. The last 15 completed scheduled runs are kept locally so you can ask about past runs in chat. "What did the morning Gmail check find?" pulls the actual saved summary, not a guess.
  
  Privacy:
  
  Your API key stays in Chrome local storage. We never see it. Page content the agent reads is sent to whichever AI provider you have selected, under their privacy terms. Nothing is sent to any server.
  
  License:
  
  The free tier covers the full core feature list. Max features unlock with a one time license. 30 day refund, no questions.

  TuClaw is a browser agent that lives in Chrome's side panel. You type a task in plain English. It clicks, types, navigates, and reports back.
  
  It works inside your real browser session. Accounts you're already signed into stay signed in. No credentials get copied anywhere new.
  
  Bring your own API key. TuClaw supports Claude (Anthropic), ChatGPT (OpenAI), and Gemini (Google). You pay your AI provider directly. No markup. Switch models from the dropdown any time.
  
  What you can do with it:
  
  Chat with your browser. Tell it what you need. It does the searching, clicking, and reading.
  
  Reference specific tabs with @. Type @ in the chat and pick the tab you want the agent to work on. Useful when your research is spread across ten tabs.
  
  Safe Mode. Before any irreversible action (submitting a form, posting publicly, deleting data) the agent stops and asks for permission. You decide.
  
  Reads the page, not pixels. The agent pulls clean text from articles, emails, and documents alongside a screenshot, so it actually understands what's on the page and adapts when sites change layout.
  
  Idle awareness. Opt in and scheduled tasks wait until you stop using the browser, so the agent never takes over your screen mid task.
  
  Persistent memory. Each session compresses into notes the agent can recall later. Relevant notes load automatically when you start a related task. The agent gets sharper at recurring work without you re-explaining context.
  
  Max features (one time purchase):
  
  Scheduled tasks. Run something every hour, every Friday at 9am, or once in 30 minutes. The agent figures out whether your prompt is a reminder ("nudge me to stretch") or a real browse task ("check Gmail and summarize what's new") and surfaces reminders as Chrome notifications.
  
  Scheduled run history. The last 15 completed scheduled runs are kept locally. You can ask "what did the morning Gmail check find?" in chat and get the saved summary back, not a guess.
  
  Privacy:
  
  Your API key stays in Chrome local storage. We never see it. Page content goes straight to your AI provider under their privacy terms. Nothing routes through our servers.
  
  License:
  
  Free tier covers the full core feature list. Max unlocks with a one time license. 30 day refund, no questions.

• May 17, 2026
  short_description

  BYOK AI agent that automates web tasks in your browser. Describe a task. AI does the clicking and typing.

  Books appointments, sends emails, scans listings, fills forms. A browser agent that lives in your side panel.

• May 17, 2026
  name

  AI Web Agent: Automate Browser Tasks

  TuClaw - Personal AI Assistant

Permissions & access

Permissions

debuggersidePanelstorageunlimitedStoragetabsalarmsnotificationsidle

Host access

<all_urls>

Screenshots