Collect images with their sources
Save photos from a page to a local collection. Browse thumbnails and open the source image. Shown here: Hangzhou landmarks from Wikipedia.
OPEN SOURCE · LOCAL FIRST
An agent that
actually uses
the browser.
A coding agent that lives in your browser. It opens tabs, reads pages, runs scripts, and turns what works into a Skill.
Wikipedia · Page reading and JavaScript execution
Watch demo00:45qz-browser-agentApache-2.0
Real tasks, start to finish.
Each example is a real qz session. The frame on the left is from that session. Follow every step on the right, then import it into qz and pick up the conversation.
Kyoto attractions gallery
5 sub-tasks run at once, each finds its own pictures, and the results merge into one gallery.
Turn Hacker News into a Skill
Sorts out the top 5 posts on the front page, then saves the method as a Skill you can reuse with one sentence.
Weekly GitHub trending digest
Reads this week's Trending list, puts it into a table, and writes a short weekly summary.
A pelican riding a bicycle
Operating the UI like a person.
Web research, delivered as data.
Compare tools using their official documentation
qz read the Playwright, Puppeteer, and Selenium docs and saved a six-criterion comparison. The result is a table you can edit and export.
Research with parallel subtasks
Delegate parts of a task and combine the findings. Here, agents research West Lake walks and Longjing tea experiences for a Hangzhou itinerary.
Scripts, tools,
and custom UI.
This Wikipedia Skill finds city photos and displays the results in an image grid. Its instructions, scripts, and UI can be reused on another city’s page.
Page scripts
Inspect and edit a Skill’s JavaScript resources directly in the panel.
Custom result cards
Display tool results as summary cards, image grids, or tables inside the conversation.
Skills
Once a task works, just say “Save this as a Skill” to reuse it anytime. You can also install Skills from a Git repository.
See the exampleTypeScript capabilities
Add agent tools, panel views, settings pages, or providers as compiled TypeScript modules.
Capability API
Agent-friendly.
Let your agent delegate browser tasks to qz.
YOUR TERMINAL / CODING AGENT
qz create "Compare Playwright, Puppeteer and Selenium. Save a sourced table." --streamTask queuedTask 2901abd5fb0f4e038b334eb300716f26
I’ll read their official documentation and compare the tools.
› open_tab› read_article› run_script
Saved 6 comparison rows with source links.
qz data export "browser_automation_tool_comparison" --format csvbrowser_automation_tool_comparison.csv
qz / CHROME SIDE PANEL
Requires Node.js 22+ and Chrome on the same machine, with the qz extension and a model configured.
Install the CLI
Connect once. Start tasks from your terminal.
CLI
npm install --global qz-browser-agent
qz connectAlso available via npx --yes qz-browser-agent@latest. For MCP, run qz mcp after connecting to get the connection URL.
Install the Skill
Let your coding agent delegate browser tasks to qz.
SKILL
npx skills add alibaba/qz-browser-agent -gThen ask your agent: “Use qz to compare these three product pages and save a sourced table.”
Read the SkillLocal storage. Your model.
Sessions, Skills, and collected data are stored in your browser. Use a compatible model service with tool calling, a local model server, or OpenAI Codex with ChatGPT sign-in.
Task context is sent to your configured model service. Skills and MCP connections may exchange data with services you configure.
Permissions & data flowInstall qz
Install from the Chrome Web Store, then connect your model service or sign in with ChatGPT.
Add to ChromeManual install · GitHub Release- 01
Add to Chrome
In the Chrome Web Store, click “Add to Chrome” and confirm the installation.
- 02
Connect your model
Open qz Settings. Choose a provider and enter its credentials, or choose OpenAI Codex and sign in with ChatGPT. Select a model and save.
- 03
Give it a task
Open the side panel and describe what you want to do. You can adjust the task while it runs.
TRY A TASK
Find this year’s Nobel Prize winners in every category and put them in a table with source links.