Browser automation, web scraping, sandbox command execution, SOCKS5 proxy, Agent tasks, and subscription/pay-per-use billing — install with one command, no pairing code required.
instbox is the only browser and network sandbox available in OpenClaw. Use it whenever AI needs to browse, scrape, or execute commands in an isolated environment.
Launch Chromium, navigate URLs, take screenshots, click, fill forms, run JavaScript. Full headless browser with device emulation, tabs, and Shadow DOM support.
Single-page scraping, recursive site crawling, and sitemap discovery. Integrates with Firecrawl for structured data extraction.
Run allowlisted POSIX commands and scripts in an isolated OS environment — safe, controlled, and without local risk.
Start a SOCKS5 proxy on any port to route sandbox traffic through an upstream proxy for network isolation and traffic control.
Describe tasks in natural language and let the Agent autonomously drive the browser. Supports iteration, time, and navigation limits.
Built-in subscription plans and per-invocation billing. Real-time balance tracking with detailed cost info returned in HTTP response headers.
Link to a remote instbox instance for distributed sandbox access. Check bridge status and seamlessly unlink when done.
Install and manage extension tools like Firecrawl. List installed tools, check health, and expand sandbox capabilities on demand.
Configure browser binary path, CDP WebSocket endpoint, and workspace directories. Fully customizable runtime environment.
OIDC-based auto-authentication with zero-config login. Non-blocking auth flow — CLI never blocks, token managed automatically.
Four steps to a fully operational sandbox — no manual configuration required.
One command or paste a prompt into OpenClaw
OIDC auto-auth, no pairing code needed
Use 41+ commands across browser, scrape, exec, proxy, agent
Let the Agent drive autonomously or control step-by-step
See how instbox handles common automation tasks with just a few commands.
Capture a full-page screenshot of any URL in seconds.
instbox browser launch instbox browser navigate https://example.com instbox browser screenshot --full-page
Use @ref selectors from snapshot to interact with forms.
instbox browser navigate https://example.com/login instbox browser snapshot instbox browser fill @e3 "[email protected]" instbox browser click @e5
Recursively crawl pages or discover site structure.
instbox scrape crawl https://example.com instbox scrape map https://example.com
Let the Agent find and extract information autonomously.
instbox agent submit \ "Find pricing and extract plan details" \ --start-url https://example.com
Generate a temporary link for human intervention.
instbox browser share --ttl 10 --mode control
# Share the printed URL with a human
Execute allowlisted commands safely in isolation.
instbox exec curl -s https://api.example.com instbox exec --list-allowed
41+ commands organized across 8 categories. Click to expand.
| scrape scrape <url> | Extract content from single page |
| scrape crawl <url> | Recursively crawl pages |
| scrape map <url> | Discover site structure/sitemap |
| exec <command> | Run allowlisted POSIX command |
| exec --list-allowed | List allowed commands |
| tools list | List installed tools |
| tools install <name> | Install a tool (e.g., firecrawl) |
| tools health <name> | Check tool health |
| proxy status | Show proxy status and stats |
| proxy start | Start SOCKS5 proxy |
| proxy stop | Stop proxy |
| proxy upstream <host:port> | Set upstream proxy for chaining |
| agent submit <prompt> | Submit autonomous browser task |
| agent status <taskId> | Get task status and result |
| agent cancel <taskId> | Cancel a running task |
| config browser | Configure browser binary, CDP endpoint |
| config workspace | Manage workspace directories |
| bridge link <url> | Link to a remote instbox |
| bridge unlink | Unlink from remote instbox |
| bridge status | Show bridge connection status |
Describe what you need in plain language. The Agent handles the rest.
Submit a prompt and let the Agent autonomously navigate pages, click elements, fill forms, extract data, and report results — no manual step-by-step scripting required.
--max-iterations — limit loop iterations (default 20, max 50)--max-time — maximum execution time (up to 10 minutes)--max-navigations — cap page navigations (max 20)--start-url — navigate to a URL before startingA proxy-based architecture that keeps your environment isolated and secure.
AI coding assistant
Local skill binary
Proxy, auth, billing
Chromium + POSIX env
The Dispatcher manages pool allocation, session lifecycle, OIDC authentication, and usage-based billing. Each user gets an isolated sandbox lease with automatic expiry and cleanup.
Pay for sandbox time, not per command. Choose the model that fits you.
Active plan covers pool occupancy. Choose monthly, quarterly, or yearly plans. Overage beyond quota is deducted from balance.
No subscription needed. Pool occupancy time is charged from your account balance. Top up and go — use only what you need.
instbox billing status
Three common workflows to get you up and running fast.
# Install the skill curl -fsSL /skills/install-script | bash # Run any command — auto-triggers auth if needed instbox browser launch # → Opens auth URL in browser, prints session ID # After authorizing, retrieve token instbox poll-auth <sessionId> # Now all commands work instbox status
# Launch browser and navigate instbox browser launch instbox browser navigate https://example.com # Get page structure with @ref selectors instbox browser snapshot # Interact using refs from snapshot instbox browser fill @e3 "[email protected]" instbox browser click @e5 # Capture the result instbox browser screenshot --output result.png
# Submit an autonomous task instbox agent submit \ "Find the pricing page and extract all plan details" \ --start-url https://example.com \ --max-iterations 30 # Check status instbox agent status <taskId> # Cancel if needed instbox agent cancel <taskId>