cmux
Drive the local cmux app — workspaces, panes, surfaces, terminal input, agent sessions, and browser surfaces.
Skill by will-ness-ai
Chalk Labs
Need help deploying cmux?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/will-ness-ai-cmux
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/cmux
What's inside
- Skill name
cmux- Description
- Drive the local cmux app — workspaces, panes, surfaces, terminal input, agent sessions, and browser surfaces.
- License
- MIT
- Instructions
- 264 lines in SKILL.md · view on GitHub
Before installing
- Instructions read files in your home folder (~/.ssh, ~/.aws…)
Files (3)
- SKILL.md
- openai.yaml · agents/
- automation-patterns.md · references/
SKILL.md (first lines)
# cmux
`cmux` is a command-line client for the cmux.app on this machine. Each command
opens the app's Unix socket and changes the live window. There is no dry-run
mode. Written against cmux 0.64.22; run `cmux version` to compare.
## The four nouns
| Noun | What it is |
|---|---|
| window | A macOS window. Holds workspaces. |
| workspace | A tab-like group in a window. Holds panes. |
| pane | A split area in a workspace. Holds surfaces. |
| surface | One tab in a pane. A terminal, a browser, a simulator, or an agent session. |
The surface is where text goes in and out. Almost every command needs one, and
takes it from a flag or from the environment.
## You are the caller
You run inside a cmux surface. Your environment has `CMUX_WORKSPACE_ID` and
`CMUX_SURFACE_ID`, and every targeting flag defaults to them. So `cmux send "ls"`
with no `--surface` types into your own terminal.
Start with:
```bash
cmux identify --json
```
`caller.surface_ref` in the output is you. `focused.surface_ref` is what the user
looks at. Send text to your own surface only when the user asks for that.
## Handles
A window, workspace, pane, or surface flag accepts three forms:
- **ref** — `workspace:30`, `surface:217`. Printed by every list command.
- **uuid** — `A1B2C3D4-E5F6-4789-ABCD-0123456789AB`. Stable and global.
- **index** — `0`, `1`, `2`. The position in the list.
A ref and an index both look like a number, and they point at different things:
the first workspace can print as `workspace:1` at index `0`. Use refs, and copy
them from a listing you just ran. Add `--id-format uuids` or `--id-format both`
when you need UUIDs.
## Orient first
```bash
cmux tree # this window: workspaces, panes, surfaces, titles, ttys
cmux tree --all # every window
cmux identify --json # who you are
```
`cmux tree` shows the whole app in one screen. Read it before you target
anything.
## Two ways in
**CLI verbs** — `cmux new-workspace`, `cmux send`, `cmux browser click`. Use
these to **create** things. A CLI verb is the only path that honours creation
options.
**`cmux rpc <method> '<json>'`** — the raw socket. Use it to **read and change
things that already exist**. It returns JSON on stdout, so it parses cleanly.
```bash
cmux rpc workspace.list
cmux rpc surface.list '{"workspace_id":"<uuid>"}'
cmux rpc surface.read_text '{"surface_id":"<uuid>"}'
cmux rpc workspace.rename '{"workspace_id":"<uuid>","title":"Done"}'
```
Creation is the split that matters. `surface.create` over rpc takes no `url`, and
workspace creation over rpc drops the title, cwd, description, and group. So:
```bash
cmux new-workspace --name "Tests" --cwd . --description "owner/repo#1" --focus false
cmux new-surface --type browser --url "file:///tmp/board.html" --workspace <uuid> --focus false
```
Errors arrive on **stderr with a non-zero exit**, not as a JSON error object:
```
Error: invalid_params: Surface is not a terminal
method_not_found: Unknown method
```
## Look it up
The binary is the source of truth, and it ships its own docs.
```bash
cmux --help # every command
cmux <command> --help # exact flags and examples for one command
cmux docs # doc topics
cmux docs api # CLI contract and socket API URLs
cmux capabilities # every v2 socket method, as JSON
```
`cmux --help` lists more than 200 commands. Do not guess a flag. Read
`cmux <command> --help` first. To find an rpc method:
```bash
cmux capabilities | jq -r '.methods[]' | grep workspace
```
Most commands take `--json` for machine-readable output.
## Stay out of the user's way
The user is sitting in this app while you work. Pass `--focus false` on creation
so their selection stays put. Focus something only when the user asks to see it.
## Recipes
### Run a command in a new workspace
```bashFrom will-ness-ai/skills, MIT.
More from will-ness-ai/skills
- 0.786score
agent-tools
will-ness-ai
Build commands for agents in AXI style.
SkillAFreshness: A★ 170Claude Code +9 - 0.742score
sandcastle
will-ness-ai
Run a Sandcastle lane — a sandboxed implement→review loop over a fixed batch of work in its own worktree, ended by an agent-authored PR.
SkillBFreshness: B★ 170Claude Code +9 - 0.739score
code-story
will-ness-ai
Build a wizard-style HTML page that teaches how and why a change works.
SkillBFreshness: B★ 170Claude Code +9 - 0.734score
grill-design
will-ness-ai
Converge on a frontend look through rounds of prototypes and grilling verdicts.
SkillBFreshness: B★ 170Claude Code +9 - 0.733score
flashlight
will-ness-ai
Shine a light into a wayfinder map's fog — work one direction now, out of frontier order, or redraw the map itself.
SkillBFreshness: B★ 170Claude Code +9 - 0.607score
test-a-skill
will-ness-ai
Field-test a skill by running it cold in parallel agent sessions, then turn what they hit into edits.
SkillBFreshness: B★ 170Claude Code +9 - 0.352score
find-standards
will-ness-ai
No description from the publisher yet.
SkillBFreshness: B★ 170Claude Code +9 - 0.349score
grilling-frontend-prototyping
will-ness-ai
No description from the publisher yet.
SkillBFreshness: B★ 170Claude Code +9
Signals
- Stars
- 170
- Forks
- 4
- Open issues
- 0
- Last commit (this folder)
- Aug 11, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/will-ness-ai/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from will-ness-ai
- 0.786score
agent-tools
will-ness-ai
Build commands for agents in AXI style.
SkillAFreshness: A★ 170Claude Code +9 - 0.742score
sandcastle
will-ness-ai
Run a Sandcastle lane — a sandboxed implement→review loop over a fixed batch of work in its own worktree, ended by an agent-authored PR.
SkillBFreshness: B★ 170Claude Code +9 - 0.739score
code-story
will-ness-ai
Build a wizard-style HTML page that teaches how and why a change works.
SkillBFreshness: B★ 170Claude Code +9 - 0.734score
grill-design
will-ness-ai
Converge on a frontend look through rounds of prototypes and grilling verdicts.
SkillBFreshness: B★ 170Claude Code +9 - 0.733score
flashlight
will-ness-ai
Shine a light into a wayfinder map's fog — work one direction now, out of frontier order, or redraw the map itself.
SkillBFreshness: B★ 170Claude Code +9 - 0.607score
test-a-skill
will-ness-ai
Field-test a skill by running it cold in parallel agent sessions, then turn what they hit into edits.
SkillBFreshness: B★ 170Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get cmux running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
cmux skill. Agent Mart. https://agentmart.dev/skills/will-ness-ai-cmux (last updated 2026-10-24)