Cmux Browser logo

Cmux Browser

Community
LeeJuOh
cmux-browser

Browser automation with cmux embedded webviews (not Chrome/Chromium). Use when $CMUX_WORKSPACE_ID is set and you need to open sites, interact with pages, wait for state changes, or extract data. Snapshot/ref workflow for reliable element targeting.

Overview

PublisherLeeJuOh
Repositoryclaude-code-zero
Skill namecmux-browser
Stars
53
Forks
4
Bundled files
9
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 9 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by LeeJuOh on GitHub. Read the source before you install it.

Installation

Install the Cmux Browser AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/LeeJuOh/claude-code-zero/tree/main/plugins/claw-mux/skills/cmux-browser
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/LeeJuOh/claude-code-zero.git /tmp/claude-code-zero
mkdir -p .claude/skills
cp -r /tmp/claude-code-zero/plugins/claw-mux/skills/cmux-browser .claude/skills/cmux-browser
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Cmux Browser in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Cmux Browser on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Cmux Browser is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Browser Automation with cmux

Use this skill for browser tasks inside cmux webviews. Requires cmux environment ($CMUX_WORKSPACE_ID). Outside cmux, this skill's features are unavailable.

Core Workflow

  1. Open or target a browser surface.
  2. Verify navigation with get url before waiting or snapshotting.
  3. Snapshot (--interactive) to get fresh element refs.
  4. Act with refs (click, fill, type, select, press).
  5. Wait for state changes.
  6. Re-snapshot after DOM/navigation changes.
bash
cmux --json browser open https://example.com
# use returned surface ref, for example: surface:7

cmux browser surface:7 get url
cmux browser surface:7 wait --load-state complete --timeout-ms 15000
cmux browser surface:7 snapshot --interactive
cmux browser surface:7 fill e1 "hello"
cmux --json browser surface:7 click e2 --snapshot-after
cmux browser surface:7 snapshot --interactive

Surface Targeting

bash
# identify current context
cmux identify --json

# open routed to a specific topology target
cmux browser open https://example.com --workspace workspace:2 --window window:1 --json

Notes:

  • CLI output defaults to short refs (surface:N, pane:N, workspace:N, window:N).
  • UUIDs are still accepted on input; only request UUID output when needed (--id-format uuids|both).
  • Keep using one surface:N per task unless you intentionally switch.

Wait Support

bash
cmux browser <surface> wait --selector "#ready" --timeout-ms 10000
cmux browser <surface> wait --text "Success" --timeout-ms 10000
cmux browser <surface> wait --url-contains "/dashboard" --timeout-ms 10000
cmux browser <surface> wait --load-state complete --timeout-ms 15000
cmux browser <surface> wait --function "document.readyState === 'complete'" --timeout-ms 10000

Common Flows

Form Submit

bash
cmux --json browser open https://example.com/signup
cmux browser surface:7 get url
cmux browser surface:7 wait --load-state complete --timeout-ms 15000
cmux browser surface:7 snapshot --interactive
cmux browser surface:7 fill e1 "Jane Doe"
cmux browser surface:7 fill e2 "jane@example.com"
cmux --json browser surface:7 click e3 --snapshot-after
cmux browser surface:7 wait --url-contains "/welcome" --timeout-ms 15000
cmux browser surface:7 snapshot --interactive

Clear an Input

bash
cmux browser surface:7 fill e11 "" --snapshot-after --json
cmux browser surface:7 get value e11 --json

Stable Agent Loop (Recommended)

bash
# navigate -> verify -> wait -> snapshot -> action -> snapshot
cmux browser surface:7 get url
cmux browser surface:7 wait --load-state complete --timeout-ms 15000
cmux browser surface:7 snapshot --interactive
cmux --json browser surface:7 click e5 --snapshot-after
cmux browser surface:7 snapshot --interactive

If get url is empty or about:blank, navigate first instead of waiting on load state.

Deep-Dive References

ReferenceWhen to Use
CommandsFull browser command mapping and quick syntax
Snapshot and RefsRef lifecycle and stale-ref troubleshooting
AuthenticationLogin/OAuth/2FA patterns and state save/load
Session ManagementMulti-surface isolation and state persistence patterns
Video RecordingRecording limitations and screenshot/snapshot alternatives
Proxy SupportProxy behavior in WKWebView and workarounds

Ready-to-Use Templates

TemplateDescription
Form AutomationSnapshot/ref form fill loop
Authenticated SessionLogin once, save/load state
Capture WorkflowNavigate + capture snapshots/screenshots

Limits (WKWebView)

These are not supported in cmux's embedded browser (WKWebView-based):

  • Viewport emulation
  • Offline emulation
  • Trace/screencast recording
  • Network route interception/mocking
  • Low-level raw input injection

Use supported high-level commands (click, fill, press, scroll, wait, snapshot) instead.

Gotchas

  • WKWebView-based, not Chromium — viewport emulation, network interception, and trace/screencast recording are not supported. Don't attempt CDP-style commands.
  • Element refs (e1, e2, ...) are invalidated on any DOM change — always re-snapshot after navigation, modal open/close, or mutating actions. Use --snapshot-after on click/fill to combine action + snapshot.
  • snapshot --interactive can return js_error on complex pages — fall back to get text body or get html body for content extraction.
  • browser open targets the workspace of the calling terminal (via $CMUX_WORKSPACE_ID), even if a different workspace is currently focused. Use --workspace to override.
  • goto and navigate are aliases — both work for same-surface navigation.
  • Always use --json with browser open — without it, the output is plain text and the returned surface ref is harder to parse reliably. cmux --json browser open <url> returns structured JSON with the surface ref.

Troubleshooting

js_error on snapshot --interactive or eval

Some complex pages can reject the JavaScript used for rich snapshots.

Recovery:

bash
cmux browser surface:7 get url
cmux browser surface:7 get text body
cmux browser surface:7 get html body
  • Use get url first to confirm the page actually navigated.
  • Fall back to get text body or get html body when snapshot --interactive returns js_error.
  • If the page is still failing, navigate to a simpler intermediate page, then retry.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Cmux Browser AI skill do?

Browser automation with cmux embedded webviews (not Chrome/Chromium). Use when $CMUX_WORKSPACE_ID is set and you need to open sites, interact with pages, wait for state changes, or extract data. Snapshot/ref workflow for reliable element targeting.

Why use Cmux Browser on TypingMind?

Because you install it once and use it with any model. Cmux Browser is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Cmux Browser in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/LeeJuOh/claude-code-zero/tree/main/plugins/claw-mux/skills/cmux-browser. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Cmux Browser?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Cmux Browser?

As many as you like. As long as a model supports skills, you can use Cmux Browser with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Cmux Browser AI skill free?

Yes. It is published on GitHub by LeeJuOh under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇