/ side panel
Read the current page
Let agents read titles, links, forms, buttons, and interactive elements from the real active tab.
Open-source AI browser assistant
SuperDuckGive AI agents your real Chrome.
SuperDuck turns the Chrome you already use into a browser tool for agents. Through the side panel, local CLI, and Agent Skill, agents can read pages, click elements, take screenshots, and inspect console or network signals while keeping your sessions, cookies, and tabs intact.
Features
What can agents do with SuperDuck?
It does not spin up a separate stateless browser. It exposes the Chrome you already use and are already logged into.
/ interaction
Click, type, and scroll
Prefer element refs for page actions, with coordinates as a fallback for complex controls or canvas UIs.
/ screenshots
Screenshots and visual checks
Capture the real browser for UI review, regression checks, and page-state verification.
/ console
Console errors
Read console errors and exceptions so agents can verify whether issues still exist after a fix.
/ network
Network requests
Inspect API requests, status codes, and failed resources to debug real page loading or backend issues.
/ agent skill
Agent workflows
Use the Skill to teach Codex, Claude Code, Cursor, and other agents the SuperDuck browser workflow.
Advantages
Why not just use a headless browser?
Many dev and ops workflows depend on real sessions, browser permissions, existing tabs, and the user's current context. That is where SuperDuck fits.
/ real session
Keeps your session
Use your own Chrome without signing internal tools, dashboards, or protected pages into another browser.
/ local bridge
Local bridge
Connect agents to the browser on your machine through the extension and native host.
/ open source
Open and inspectable
The extension, native host, CLI, and Skill live on GitHub for review, customization, and self-hosting.
Install
Connect SuperDuck in three steps.
Order matters: install the Chrome side panel first, then the local CLI bridge, and add the Agent Skill last.
Step 1
Install the Chrome side panel
The extension provides the side panel service, tab control, debugger access, and Native Messaging entry.
Step 2
Install the local CLI bridge
The CLI passes agent requests to the native host, then into the Chrome extension.
Step 3
Install the Agent Skill last
The Skill teaches agents how to use SuperDuck's read, click, screenshot, and diagnostics workflows.
Commands
The common commands stay small.
After the Chrome extension is installed, CLI handles setup, doctor, and browser control; the Skill teaches agents how to use it.
superduck setuplocal
# 1. Add the Chrome side panel from Web Store first
$ npm install -g superduck-cli
$ superduck setup
$ superduck doctor
✓ extension responds
✓ native-host UDS reachable
$ npx skills add superduck-ai/superduck
npm install -g superduck-cli
npx skills add superduck-ai/superduck
superduck doctor
GitHub
Source code, Skill, and packages.
Find SuperDuck's Chrome extension, native host, CLI, npm package, and Skill from these links.
GitHub
Source repository
Chrome extension, native host, CLI, Skill, and docs.
↗
Chrome
Chrome side panel extension
Install the browser side panel service from Chrome Web Store.
↗
Skill
SuperDuck Agent Skill
Teach Codex, Claude Code, Cursor, and other agents the SuperDuck workflow.
↗
npm
superduck-cli
Native binary distribution and CLI setup.
↗
Let agents use the browser you are already logged into.
Start with the Chrome side panel, connect the CLI, then add the SuperDuck Skill to your agent.