New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Waku Computer Use

by egoist in egoist/waku

Control local macOS, Windows, and Linux apps through Waku Computer Use.

GPL-3.0Auto-check passedProductivity & Automation

Install Waku Computer Use

skills CLI
$ npx skills add egoist/waku --skill waku-computer-use -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install egoist/waku waku-computer-use --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/egoist/waku.git skills-src && mkdir -p .claude/skills && cp -r skills-src/resources/computer-use .claude/skills/waku-computer-use && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
waku-computer-use
GitHub stars
1.6k
Token cost
~3.6k tokens
SKILL.md length
404 words
Files
8
Skills in repo
1
Repo updated
First seen
Licence
GPL-3.0

At a glance

Control local macOS, Windows, and Linux apps through Waku Computer Use.

  • Tasks that involve Desktop control
  • SKILL.md covers Bootstrap, API surface, Observe an app and Act and verify
  • Runs Swift, Rust and TypeScript scripts from its folder

What it does

Waku Computer Use is an agent skill from egoist/waku. Control local macOS, Windows, and Linux apps through Waku Computer Use. Prefer purpose-built connectors, APIs, or CLIs when available.

Its SKILL.md is about 3.6k tokens, which your agent loads only when the skill is triggered. The skill folder holds 7 other files (for example `pi-extension.ts`).

It sits in Productivity & Automation, covering Desktop control. It works with Linux, macOS and Rust. The repository describes itself as: ⚡ A native app for all your coding agents. The licence is GPL-3.0.

When your agent uses it

  • Tasks that involve Desktop control

Example prompts

  • “/waku-computer-use”

Requirements

  • Node.js

What it can do on your machine

Read from SKILL.md and the folder at commit 10bcd72. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships script files (Swift, Rust and TypeScript), which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Waku Computer Use loads about 3.6k tokens when it runs. Until then it costs about 38 tokens; SKILL.md has 404 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~38
When it runs · the whole SKILL.md, loaded when a task matches
~3.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from egoist/waku at commit 10bcd72, republished under its GPL-3.0 licence (© egoist). 404 words, ~3,632 tokens.

Download SKILL.mdSave it as .claude/skills/waku-computer-use/SKILL.md (or your agent's skills folder). This skill also uses 7 other files; get the full folder from GitHub.
name
waku-computer-use
description
Control local macOS, Windows, and Linux apps through Waku Computer Use. Prefer purpose-built connectors, APIs, or CLIs when available.

Waku Computer Use

Use the js tool from waku_js_repl for computer interactions. It runs a persistent QuickJS kernel. Use the direct cua methods documented here; all methods are available immediately after bootstrap. Waku shows Cua's native virtual cursor automatically while actions run; no cursor setup is needed.

Bootstrap

Run once per fresh JavaScript session:

js
if (!globalThis.cua) {
  await setupComputerUseRuntime({ globals: globalThis });
}

After js_reset, bootstrap again before using cua. Module imports and Node subprocess APIs are unavailable. The native runtime is managed by Waku.

Use jsRepl.write(value) for text or structured output and await jsRepl.emitImage(image) to show a returned image. Prefer top-level var for names reused across calls. Calls default to 30 seconds; set timeout_ms when an observation needs longer.

<!-- BEGIN NATIVE API -->

API surface

Call these methods directly. Arguments use the native names shown below; conditional constraints are validated by the runtime.

ts
type CuaResult = {
  content: Array<{ type: 'text'; text: string } | { type: 'image'; data: string; mimeType: string }>;
  structuredContent?: any;
  isError?: boolean;
};

declare const cua: {
  readonly platform: "macos";
  list_apps(args?: Record<string, never>): Promise<CuaResult>;
  list_windows(args?: { on_screen_only?: boolean; pid?: number }): Promise<CuaResult>;
  get_window_state(args: { capture_mode?: "ax" | "vision"; include_accessibility_tree?: boolean; include_screenshot?: boolean; max_depth?: number; max_dimension?: number; max_elements?: number; pid: number; query?: string; screenshot_out_file?: string; session?: string; window_id: number }): Promise<CuaResult>;
  verify_state(args: { expect: Array<{ element?: { enabled?: boolean | null; exists?: true; selected?: boolean | null; selector: { label_contains?: string; role?: string }; value_equals?: string | null } | null; window?: { bounds?: { height: number; tolerance_px?: number; width: number; x: number; y: number } | null; exists?: boolean | null } | null }>; include_screenshot?: boolean | null; pid: number; session?: string; stable_samples?: number; timeout_ms?: number; window_id: number }): Promise<CuaResult>;
  launch_app(args?: { additional_arguments?: Array<string>; bundle_id?: string; creates_new_application_instance?: boolean; name?: string; urls?: Array<string>; webkit_inspector_port?: number }): Promise<CuaResult>;
  kill_app(args: { pid: number }): Promise<CuaResult>;
  set_window_frame(args: { height: number; pid: number; session?: string; width: number; window_id: number; x: number; y: number }): Promise<CuaResult>;
  invoke_menu(args: { path: Array<string>; pid: number; session?: string; window_id: number }): Promise<CuaResult>;
  click(args?: { action?: string; button?: "left" | "right" | "middle"; count?: number; debug_image_out?: string; delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; from_zoom?: boolean; modifier?: Array<string>; pid?: number; scope?: "window" | "desktop"; session?: string; snapshot_id?: string; target?: { kind: "window"; pid: number; window_id: number; [key: string]: unknown } | { display_id: string; kind: "desktop"; [key: string]: unknown }; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  double_click(args: { delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; pid: number; session?: string; snapshot_id?: string; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  right_click(args: { delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; modifier?: Array<string>; pid: number; session?: string; snapshot_id?: string; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  drag(args: { button?: "left" | "right" | "middle"; delivery_mode?: "background" | "foreground"; duration_ms?: number; from_x: number; from_y: number; from_zoom?: boolean; modifier?: Array<string>; pid?: number; scope?: "window" | "desktop"; session?: string; steps?: number; target?: null; to_x: number; to_y: number; window_id?: number }): Promise<CuaResult>;
  type_text(args: { delay_ms?: number; delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; pid?: number; scope?: "window" | "desktop"; session?: string; snapshot_id?: string; target?: null; text: string; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  press_key(args: { delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; key: string; modifiers?: Array<string>; pid?: number; scope?: "window" | "desktop"; session?: string; snapshot_id?: string; target?: null; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  hotkey(args: { delivery_mode?: "background" | "foreground"; element_index?: number; element_token?: string; keys: Array<string>; pid?: number; scope?: "window" | "desktop"; session?: string; snapshot_id?: string; target?: null; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  set_value(args: { element_index?: number; element_token?: string; pid: number; session?: string; snapshot_id?: string; value: string; window_id?: number }): Promise<CuaResult>;
  scroll(args: { amount?: number; by?: "line" | "page"; delivery_mode?: "background" | "foreground"; direction: "up" | "down" | "left" | "right"; element_index?: number; element_token?: string; pid?: number; scope?: "window" | "desktop"; session?: string; snapshot_id?: string; target?: null; window_id?: number; x?: number; y?: number }): Promise<CuaResult>;
  clipboard_read(args?: { include_text?: boolean; session?: string }): Promise<CuaResult>;
  clipboard_write(args?: { file_path?: string; image_path?: string; session?: string; text?: string }): Promise<CuaResult>;
  get_screen_size(args?: { session?: string }): Promise<CuaResult>;
  get_desktop_state(args?: { screenshot_out_file?: string; session?: string }): Promise<CuaResult>;
  get_cursor_position(args?: { session?: string }): Promise<CuaResult>;
  move_cursor(args: { cursor_id?: string; scope?: "window" | "desktop"; session?: string; target?: null; x: number; y: number }): Promise<CuaResult>;
  set_agent_cursor_enabled(args: { enabled: boolean; session: string }): Promise<CuaResult>;
  set_agent_cursor_motion(args: { arc_flow?: number | null; arc_size?: number | null; dwell_after_click_ms?: number | null; end_handle?: number | null; glide_duration_ms?: number | null; idle_hide_ms?: number | null; session: string; spring?: number | null; start_handle?: number | null; turn_radius?: number | null }): Promise<CuaResult>;
  set_agent_cursor_theme(args: { reduced_motion?: "auto" | "on" | "off"; session: string; theme_id: string }): Promise<CuaResult>;
  get_agent_cursor_state(args: { session: string }): Promise<CuaResult>;
  check_permissions(args?: { probe_direct_capture?: boolean; prompt?: boolean }): Promise<CuaResult>;
  health_report(args?: { include?: Array<string>; skip?: Array<string> }): Promise<CuaResult>;
  get_config(args?: Record<string, never>): Promise<CuaResult>;
  set_config(args?: { experimental_pip?: boolean; experimental_pip_geometry?: string; key?: string; max_image_dimension?: number; value?: unknown }): Promise<CuaResult>;
  get_accessibility_tree(args?: Record<string, never>): Promise<CuaResult>;
  zoom(args: { pid?: number; window_id: number; x1: number; x2: number; y1: number; y2: number }): Promise<CuaResult>;
  page(args: { action: "execute_javascript" | "get_text" | "query_dom" | "click_element" | "insert_text" | "type_keystrokes" | "enable_javascript_apple_events"; attributes?: Array<string>; bundle_id?: string; cdp_port?: number; css_selector?: string; javascript?: string; pid?: number; selector?: string; target_url_contains?: string; text?: string; user_has_confirmed_enabling?: boolean; window_id?: number }): Promise<CuaResult>;
  get_browser_state(args?: { continuation?: string; include_screenshot?: boolean; pid?: number; query?: string; scope_ref?: string; session?: string; snapshot_format?: "dom_refs_v1" | "semantic_v2"; tab_id?: string; target_id?: string; window_id?: number; [key: string]: unknown }): Promise<CuaResult>;
  browser_prepare(args?: { allow_launch?: boolean; pid?: number; profile?: { mode: "isolated_new" | "isolated_named"; name?: string }; session?: string; strategy?: { kind: "existing_profile" }; window_id?: number; [key: string]: unknown }): Promise<CuaResult>;
  browser_navigate(args: { session?: string; tab_id: string; target_id: string; url: string; [key: string]: unknown }): Promise<CuaResult>;
  browser_click(args: { input_route?: "trusted" | "dom_event"; ref?: string; session?: string; tab_id: string; target_id: string; x?: number; y?: number; [key: string]: unknown }): Promise<CuaResult>;
  browser_type(args: { mode?: "insert_text" | "keystrokes"; ref: string; replace?: boolean; session?: string; tab_id: string; target_id: string; text: string; [key: string]: unknown }): Promise<CuaResult>;
  browser_dialog(args: { action: "inspect" | "accept" | "dismiss"; delivery_mode?: "background" | "foreground"; dialog_id?: string; prompt_text?: string; session?: string; tab_id: string; target_id: string; [key: string]: unknown }): Promise<CuaResult>;
  browser_set_input_files(args: { files: Array<string>; ref: string; session?: string; tab_id: string; target_id: string; [key: string]: unknown }): Promise<CuaResult>;
  browser_download(args: { destination_root: string; ref: string; session: string; tab_id: string; target_id: string; [key: string]: unknown }): Promise<CuaResult>;
  browser_pointer(args: { action: "hover" | "right_click" | "double_click" | "scroll" | "drag"; delta_x?: number; delta_y?: number; destination_ref?: string; input_route?: "trusted" | "dom_event"; ref?: string; session: string; tab_id: string; target_id: string; to_x?: number; to_y?: number; x?: number; y?: number; [key: string]: unknown }): Promise<CuaResult>;
  start_recording(args: { output_dir: string; record_video?: boolean }): Promise<CuaResult>;
  stop_recording(args?: Record<string, never>): Promise<CuaResult>;
  get_recording_state(args?: Record<string, never>): Promise<CuaResult>;
  replay_trajectory(args: { delay_ms?: number; dir: string; stop_on_error?: boolean }): Promise<CuaResult>;
  install_ffmpeg(args?: { confirm?: boolean }): Promise<CuaResult>;
  start_session(args?: { capture_scope?: "auto" | "window" | "desktop"; cursor_theme?: { reduced_motion?: "auto" | "on" | "off"; theme_id: string; [key: string]: unknown } | null; session?: string; [key: string]: unknown }): Promise<CuaResult>;
  escalate_session(args: { detail?: string; reason: "ax_tree_pixel_mismatch" | "background_delivery_failed" | "foreground_ineffective" | "no_window_target" | "other"; session: string; [key: string]: unknown }): Promise<CuaResult>;
  get_session(args?: { session?: string; [key: string]: unknown }): Promise<CuaResult>;
  list_sessions(args?: { cursor?: string; limit?: number | null; [key: string]: unknown }): Promise<CuaResult>;
  get_session_state(args?: { session?: string; [key: string]: unknown }): Promise<CuaResult>;
  end_session(args?: { session?: string; [key: string]: unknown }): Promise<CuaResult>;
};
<!-- END NATIVE API -->

Observe an app

Use the app and window identities already known from the task or a previous observation. Otherwise, list apps directly and select the requested app:

js
var appList = await cua.list_apps();
jsRepl.write(appList.structuredContent.apps);

Select the intended app by its observed name, bundle ID, or launch path. If it is not running, use cua.launch_app(...) with the app's observed identity. List its windows and select the intended title:

js
// appInfo is the selected entry from appList.structuredContent.apps.
var windows = await cua.list_windows({ pid: appInfo.pid });
jsRepl.write(windows.structuredContent.windows);
js
// windowInfo is selected from windows.structuredContent.windows.
var captureArgs = { pid: appInfo.pid, window_id: windowInfo.window_id };
var state = await cua.get_window_state(captureArgs);
jsRepl.write(state.structuredContent);
for (var image of state.content ?? []) {
  if (image.type === "image") await jsRepl.emitImage(image);
}

Each method returns content, optional structuredContent, and optional isError. If isError is true, read the error before deciding what to do next. Keep native window IDs intact; do not convert them to 32-bit integers.

Show full SKILL.md (179 more words)Show less

Act and verify

Use an element_token from the latest window state when available. For pixel actions, use coordinates from that exact window screenshot. Cua handles backing scale; do not resize the image or add the window's screen origin.

js
// button is the intended element from state.structuredContent.elements.
var result = await cua.click({
  pid: appInfo.pid,
  window_id: windowInfo.window_id,
  element_token: button.element_token,
  delivery_mode: "background",
});
jsRepl.write(result);
var after = await cua.get_window_state(captureArgs);
jsRepl.write(after.structuredContent);

Typing and key presses use the same observed window:

js
await cua.type_text({ pid: appInfo.pid, window_id: windowInfo.window_id, text: "hello" });
await cua.press_key({ pid: appInfo.pid, window_id: windowInfo.window_id, key: "return" });

Confirm the intended effect from a fresh state after acting. For asynchronous changes, poll observations with a bounded deadline without repeating the input. Tokens expire after a new snapshot, window, or session. Refresh stale tokens. A truncated or degraded tree may omit controls; use its screenshot and the returned capability information when choosing another route.

Prefer background delivery. An unsupported route does not authorize an automatic foreground retry. Windows secure/elevated desktops and Linux compositor restrictions remain in effect. cua.check_permissions() reports capabilities; macOS permission prompts belong to Waku Settings > Computer Use.

Stop, reset, and disconnect end the helper's work. An interrupted action may have completed, so inspect fresh state before retrying it. One implicit session is reused across calls. If using a named session, repeat that label on every call that accepts it.

© egoist, GPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 7 other files in resources/computer-use of egoist/waku.

  • SKILL.md
  • CUA-LICENSE
  • CuaDriver.swift
  • Info.plist
  • WakuComputerUse.swift
  • cua-host.h
  • cua-host.rs
  • pi-extension.ts

Open the folder on GitHubat commit 10bcd72

Compare with similar skills

Waku Computer Use next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Waku Computer Use compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Waku Computer Use this skillegoist/waku1.6k—~3.6kAutomated safety check: PassGPL-3.0
Open Computer UseiFurySt/open-codex-computer-use2.4k—~1.5kAutomated safety check: PassMIT
Linux Desktop Controlagent-sh/computer-use-linux669—~2.8kAutomated safety check: PassMIT
Drive Screencoleam00/skills674—~5.4kAutomated safety check: PassMIT
Browser MCP Agentantibrow/anti-detect-browser-skills171 repos~4.2kAutomated safety check: WarnMIT
Gooeypi Computer Useam-will/gooey-pi941—~370Automated safety check: PassMIT

Similar skills

  • Open Computer Use

    iFurySt/open-codex-computer-use

    Platform-neutral guidance for using Open Computer Use, the open-source Computer Use MCP server and CLI for macOS, Linux, and Windows.

    2.4k GitHub stars~1.5k tokensUpdated 3 days ago
    Productivity & AutomationAuto-check passed

  • Linux Desktop Control

    agent-sh/computer-use-linux

    Lets an agent observe and operate a local Linux desktop through the computer-use-linux MCP server or Pi tools: accessibility trees, screenshots, windows and input.

    669 GitHub stars~2.8k tokensUpdated yesterday
    Productivity & AutomationAuto-check passed

  • Drive Screen

    coleam00/skills

    Take real control of the desktop - list and focus windows, type, paste, click, scroll, and screenshot - on Windows, macOS or Linux, and drive other coding-agent sessions running in terminals.

    674 GitHub stars~5.4k tokensUpdated 2 days ago
    Productivity & AutomationAuto-check passed

  • Browser MCP Agent

    antibrow/anti-detect-browser-skills

    Give an AI agent its own real browser over MCP tool calls - launch, navigate, click, fill, screenshot, extract text, run JS - with a kernel-level real-device fingerprint and a persistent profile, so…

    17 GitHub starsUsed in 1 repo~4.2k tokens
    Productivity & AutomationAuto-check: warnings

  • Gooeypi Computer Use

    am-will/gooey-pi

    Drive native applications on macOS, Windows, or Linux through the separately installed TryCUA driver CLI.

    941 GitHub stars~370 tokensUpdated 1 mo ago
    Productivity & AutomationAuto-check passed

  • Computer Use

    johnson7788/MultiUserClaw

    Drive the user's desktop in the background — clicking, typing, scrolling, dragging — without stealing the cursor, keyboard focus, or switching virtual desktops / Spaces.

    327 GitHub starsUsed in 1 repo~2.8k tokens
    Productivity & AutomationAuto-check: warnings

Works with

Categories

Questions about Waku Computer Use

What does Waku Computer Use do?

Control local macOS, Windows, and Linux apps through Waku Computer Use. Waku Computer Use is an agent skill from egoist/waku. Control local macOS, Windows, and Linux apps through Waku Computer Use.

When should I use Waku Computer Use?

Waku Computer Use fits situations like: tasks that involve Desktop control.

How do I install Waku Computer Use in Claude Code?

Run `npx skills add egoist/waku --skill waku-computer-use -a claude-code`. Or copy the skill folder (resources/computer-use in egoist/waku) into .claude/skills/waku-computer-use in your project. Claude Code loads it when a task matches its description.

How do I install Waku Computer Use in Codex?

Run `npx skills add egoist/waku --skill waku-computer-use -a codex`. Or copy the skill folder (resources/computer-use in egoist/waku) into .agents/skills/waku-computer-use in your project. Codex loads it when a task matches its description.

Can I use Waku Computer Use in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add egoist/waku --skill waku-computer-use -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/waku-computer-use, .gemini/skills/waku-computer-use, .github/skills/waku-computer-use and .opencode/skills/waku-computer-use in your project.

What does Waku Computer Use need to run?

Going by SKILL.md and its folder, Waku Computer Use needs Swift, Rust and TypeScript for the scripts in its folder. Our summary lists: Node.js.

Does Waku Computer Use access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Waku Computer Use safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Waku Computer Use use?

Waku Computer Use is published under the GPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Waku Computer Use use?

About 3.6k tokens (SKILL.md is roughly 15k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Waku Computer Use?

Skills that share tags, products or a category with Waku Computer Use: Open Computer Use (iFurySt/open-codex-computer-use, 2.4k stars), Linux Desktop Control (agent-sh/computer-use-linux, 669 stars), Drive Screen (coleam00/skills, 674 stars) and Browser MCP Agent (antibrow/anti-detect-browser-skills, 17 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Waku Computer Use?

egoist (a GitHub user) maintains it in egoist/waku, which has 1,567 GitHub stars. The repository was last updated on October 2, 2026.

Source: egoist/waku on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.