terminal

Browser Use Terminal

Automate the boring stuff in the browser.

Browser Use Terminal is a Rust TUI for browser agents. It combines a new LLM harness, Browser Harness-style CDP control, real Chrome sessions, and a terminal UI you can actually steer.

curl -fsSL https://browser-use.com/terminal/install.sh | sh
browser

What It Does

• runs browser tasks from a terminal UI
• works with your logged-in Chrome when the task needs real account state
• supports headless Chromium and Browser Use cloud for clean or remote runs
• lets you watch, steer, stop, retry, and resume tasks
• keeps local history, screenshots, artifacts, and follow-ups
• uses a new LLM harness built to be 2x cheaper and 2x faster than Browser Harness

How It Works

Browser Use Terminal is a browser-first LLM harness: Rust owns the agent loop and durable state, while the browser runtime gives the model direct CDP control over Chrome.

you
 │
 ▼
browser terminal
 │
 ├─ custom Ratatui UI      watch · steer · stop · resume
 ├─ Rust LLM harness       tools · subagents · compaction · cancellation
 ├─ SQLite event log       history · screenshots · artifacts · traces
 └─ CDP browser runtime    profiles · doctor · recovery · ownership
      │
      ▼
 real Chrome  |  headless Chromium  |  Browser Use cloud

• The model gets raw browser capability: CDP, page JavaScript, screenshots, files, and helper code.
• The terminal stays controllable: history, artifacts, follow-ups, browser recovery, and a custom Ratatui renderer for live work.

Try It

Get my San Francisco parking permit.

Give this employee admin permission in Azure.

Find the cancellation policy for my current hotel reservation.

Setup

Launch the app:

browser

Use slash commands inside the TUI:

/auth      sign in
/model     choose a model
/browser   choose local, headless, or cloud browser
/update    update the app

Useful shell commands:

browser auth status
browser config show
browser diagnostics

Use It From Claude Code, Codex, or OpenCode

Browser Use Terminal plugs into any coding assistant that can run shell commands, browser-harness style: a skill teaches the assistant the CLI, and the CLI hands it the whole browser runtime.

browser-use-terminal skill install        # registers the skill for detected assistants

Then ask your assistant to browse:

browser-use-terminal browser exec <<'PY'
new_tab("https://example.com")
wait_for_load()
print(capture_screenshot())
PY

Screenshots are saved as files and the path is printed, so assistants view them with their native file-reading tools (Claude Code Read, Codex view_image, OpenCode read). The browser persists between calls. See docs/assistant-plugins.md.

Development

cargo fmt --check
cargo test
uv run --with pytest python -m pytest -q
scripts/verify-terminal-ui.sh

Terminal UI changes must pass the full verification script. It runs Rust tests, Python tests, deterministic Ratatui dumps, and a real tmux smoke test.

Agent Tools

Release archives include a managed ripgrep binary at bin/agent-tools/rg. Agent shell commands prepend Browser Use Terminal's temporary tool shim and managed tools directory to PATH, so rg, rg --version, and rg --files work without depending on the user's shell dotfiles or DotSlash. Source checkout entry points and local dev wrappers install the same managed ripgrep under target/debug/agent-tools/rg; run scripts/install-agent-ripgrep.sh target/debug/agent-tools to refresh it manually.

Telemetry

You can disable (100% completely anonymous) telemetry with BUT_TELEMETRY=0.

Docs

• docs/terminal-ui-product-ux.md
• docs/terminal-ui-testing.md
• docs/terminal-renderer-architecture.md
• docs/assistant-plugins.md

License

MIT