Mac Computer Use
To3akaRin/mac-computer-use
操作 macOS 桌面应用,探测窗口和自动化接口、截图、读取或修改辅助功能元素、执行鼠标键盘动作,以及通过 CDP 操作内嵌 Chromium 页面。适用于桌面应用自动化与界面验收;普通网页任务优先使用已有浏览器工具。
Verify Reel, the macOS menu-bar screen recorder, by launching a disposable app bundle and driving its real UI with Computer Use.
$ npx skills add rselbach/reel --skill verify-reel -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install rselbach/reel verify-reel --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/verify-reel .claude/skills/verify-reel && rm -rf skills-srcUse ~/.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/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .claude/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reelType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add rselbach/reel --skill verify-reel -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install rselbach/reel verify-reel --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.agents/skills/verify-reel .agents/skills/verify-reel && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .agents/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add rselbach/reel --skill verify-reel -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install rselbach/reel verify-reel --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.agents/skills/verify-reel .cursor/skills/verify-reel && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .cursor/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/rselbach/reel.git --path .agents/skills/verify-reel--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add rselbach/reel --skill verify-reel -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install rselbach/reel verify-reel --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.agents/skills/verify-reel .gemini/skills/verify-reel && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .gemini/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install rselbach/reel verify-reelInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add rselbach/reel --skill verify-reel -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .github/skills && cp -r skills-src/.agents/skills/verify-reel .github/skills/verify-reel && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .github/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add rselbach/reel --skill verify-reel -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install rselbach/reel verify-reel --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rselbach/reel.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.agents/skills/verify-reel .opencode/skills/verify-reel && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "verify-reel" agent skill from https://github.com/rselbach/reel/tree/main/.agents/skills/verify-reel into .opencode/skills/verify-reel/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "verify-reel", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
verify-reelVerify Reel, the macOS menu-bar screen recorder, by launching a disposable app bundle and driving its real UI with Computer Use.
Verify Reel is an agent skill from rselbach/reel. Verify Reel, the macOS menu-bar screen recorder, by launching a disposable app bundle and driving its real UI with Computer Use. Use after changes to first-run, recording, settings, or preview behavior.
Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 9 other files, including scripts (for example `features/README.md`, `features/choose-recording-target.md` and `features/first-launch.md`).
It sits in Productivity & Automation, covering Desktop control. It works with macOS. The licence is Unlicense.
Read from SKILL.md and the folder at commit b24b76a. It shows what the files ask for, not the result of running them.
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.
Ships 1 file in scripts/, which the agent can run.
Shell commands in SKILL.md call:
justFrom the folder's file list and the shell code blocks in SKILL.md.
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Verify Reel loads about 2k tokens when it runs. Until then it costs about 54 tokens; SKILL.md has 825 words of instructions outside code blocks.
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.
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); the scripts in this folder are not scanned.
The full file from rselbach/reel at commit b24b76a, republished under its Unlicense licence (© rselbach). 825 words, ~2,032 tokens.
.claude/skills/verify-reel/SKILL.md (or your agent's skills folder). This skill also uses 7 other files; get the full folder from GitHub.Drive the packaged Reel app as a user would. Read
features/README.md first, then read the file for the
feature under test.
Reel is a menu-bar macOS app. It has no CLI or HTTP interface. Use the
computer-use:computer-use skill and its node_repl workflow for every UI
action. Shell commands may build, launch, inspect, and stop the disposable
instance.
Start from the repository root. The helper refuses to launch if any Reel process is already running because Reel registers machine-wide hotkeys and uses machine-wide screen, microphone, and camera permissions.
run_id="$(date -u +%Y%m%dT%H%M%SZ)-$$"
.agents/skills/verify-reel/scripts/verify-reel launch "${run_id}"To re-sign the disposable copy after its bundle metadata changes, pass a Developer ID Application identity:
SIGNING_IDENTITY="<Developer ID Application identity>" \
.agents/skills/verify-reel/scripts/verify-reel launch "${run_id}"launch runs just build-app, copies the result to
.build/verify-reel/Reel Verification.app, changes only the copied bundle's
identifier and display name, and starts its executable. It sets
CFFIXED_USER_HOME to the run directory, so preferences and the default Movies
folder do not touch Reel's normal data.
When SIGNING_IDENTITY is set, launch signs Sparkle.framework, the Reel
executable with Reel.entitlements, and the copied bundle in that order. The
helper verifies the copied bundle's signature before launch. It does not change
.build/Reel.app.
The app is ready when launch reports ready, doctor passes, and Computer
Use can read either the Welcome to Reel window or Reel's menu-bar item. The
first launch does not request screen-recording permission until the user chooses
Grant Screen Recording Access.
Do not start a second run. If launch reports another Reel process, leave that process alone and ask the user to quit it.
macOS applies a new screen-recording grant only after the requesting app quits. Keep the current run and its signed bundle. After granting access, relaunch it without rebuilding or signing again:
.agents/skills/verify-reel/scripts/verify-reel relaunch "${run_id}"relaunch stops the recorded process if it is still alive, then opens the same
bundle with the same isolated home. If Reel already relaunched itself, the
helper adopts the replacement process instead. It updates the recorded PID and
runs doctor before returning. Do not run cleanup between the permission
grant and this relaunch.
Run this read-only check before driving the app and whenever the UI looks stale:
.agents/skills/verify-reel/scripts/verify-reel doctor "${run_id}"Doctor requires the recorded PID to be alive, its command to match the copied
executable, the copied bundle ID to be com.rselbach.reel.verification, and the
embedded commit to match HEAD. It prints the app path, PID, bundle ID,
evidence directory, and current network-connection count. A failed doctor means
the instance is not safe to drive. Clean it up, then launch a new run.
Load the computer-use:computer-use skill. Initialize its bundled package once
in a fresh node_repl session:
globalThis.sky = (await import("@oai/sky")).sky;
globalThis.reelFs = await import("node:fs/promises");
globalThis.reelFileURLToPath = (await import("node:url")).fileURLToPath;
globalThis.reelRepo = nodeRepl.cwd;
globalThis.reelRunId = (
await reelFs.readFile(
`${reelRepo}/.build/verify-reel/current-run`,
"utf8"
)
).trim();
globalThis.reelRunDir =
`${reelRepo}/.build/verify-reel/runs/${reelRunId}`;
globalThis.reelAppPath = (
await reelFs.readFile(`${reelRunDir}/app-path`, "utf8")
).trim();
globalThis.reelEvidenceDir = (
await reelFs.readFile(`${reelRunDir}/evidence-dir`, "utf8")
).trim();
globalThis.captureReel = async (name) => {
const state = await sky.get_app_state({
app: reelAppPath,
disableDiff: true,
});
await reelFs.mkdir(reelEvidenceDir, { recursive: true });
await reelFs.writeFile(
`${reelEvidenceDir}/${name}.ax.txt`,
`${state.text}\n`
);
if (!state.screenshot) {
throw new Error(`Reel returned no screenshot for ${name}`);
}
await reelFs.copyFile(
reelFileURLToPath(state.screenshot.url),
`${reelEvidenceDir}/${name}.png`
);
return state;
};
globalThis.reelState = await captureReel("initial");
nodeRepl.write(reelState.text);Use the full app path from the run metadata. Do not target Reel by display
name because a normal installation may exist on the same Mac.
Derive every element_index from the latest accessibility tree. After each
click, key press, drag, or value change, call get_app_state again before
choosing the next index. Stable handles in this app include:
Welcome to Reel, New Recording, Reel Settings,
Recording Preview, and About ReelGet Started, Start Recording...,
Settings..., Stop Recording, Record Again, and DoneGeneral, Recording, and ShortcutsReelUse coordinates only when the accessibility tree omits a visual control, such as a trim handle or a recording-region corner. Capture a fresh screenshot before choosing coordinates and another one after the action.
Proof belongs in .build/verification-evidence/reel/<run-id>/. The capture
helper writes matching .ax.txt and .png files there. Name captures after the
feature and state, for example settings-before and settings-after.
A valid proof must:
For a preference side effect, capture the helper's read-only result:
.agents/skills/verify-reel/scripts/verify-reel preference \
"${run_id}" hasShownWelcome \
> ".build/verification-evidence/reel/${run_id}/hasShownWelcome.txt"For a recording, capture the picker, the recording state, and the preview.
Also record stat and mdls output for the MP4 under the run's isolated Movies
directory. Confirm that dry-run or safe-mode claims match observed files,
network connections, and UI behavior. Do not infer safety from a label.
Stop only the recorded PID, then remove the copied app and per-run state:
.agents/skills/verify-reel/scripts/verify-reel cleanup "${run_id}"Cleanup verifies that the PID still belongs to the exact executable before it
sends a signal. It never kills by process name. It removes
.build/verify-reel/Reel Verification.app and that run's scratch directory.
It does not remove .build/verification-evidence/reel/<run-id>/.
After cleanup, prove that the evidence survived:
test -d ".build/verification-evidence/reel/${run_id}"
find ".build/verification-evidence/reel/${run_id}" -type f -maxdepth 1 -printIf a drive fails, run cleanup before retrying. If cleanup refuses because the PID now names another executable, do not signal it. Report the mismatch.
The executable helper is
.agents/skills/verify-reel/scripts/verify-reel.
verify-reel launch <run-id>
verify-reel relaunch <run-id>
verify-reel doctor <run-id>
verify-reel describe <run-id>
verify-reel preference <run-id> <key>
verify-reel cleanup <run-id>describe prints the recorded paths and identity. preference reads one key
from the run's isolated defaults domain. Neither command changes app state.
© rselbach, Unlicense. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 7 other files (scripts) in .agents/skills/verify-reel of rselbach/reel.
Open the folder on GitHubat commit b24b76a
Verify Reel 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.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Verify Reel this skillrselbach/reel | 118 | — | ~2k | Automated safety check: Pass | Unlicense | |
| Mac Computer UseTo3akaRin/mac-computer-use | 1.1k | — | ~495 | Automated safety check: Pass | MIT | |
| Steerdisler/mac-mini-agent | 272 | — | ~2.1k | Automated safety check: Pass | None | |
| Computer Usekunpengtalk/PeakCode | 166 | — | ~2.5k | Automated safety check: Pass | MIT | |
| Openbridge DebugAFK-surf/OpenBridge | 430 | — | ~2.2k | Automated safety check: Pass | MIT | |
| Computer Useclacky-ai/openclacky | 1.2k | — | ~1.7k | Automated safety check: Pass | MIT |
To3akaRin/mac-computer-use
操作 macOS 桌面应用,探测窗口和自动化接口、截图、读取或修改辅助功能元素、执行鼠标键盘动作,以及通过 CDP 操作内嵌 Chromium 页面。适用于桌面应用自动化与界面验收;普通网页任务优先使用已有浏览器工具。
disler/mac-mini-agent
macOS GUI automation CLI. An agent skill from disler/mac-mini-agent.
kunpengtalk/PeakCode
Drive a macOS app through the computer tool — read its accessibility tree, act on elements, and use real input only when the tree cannot reach the target.
AFK-surf/OpenBridge
Debug and validate OpenBridge on a real macOS desktop through mini-machine plus full VNC computer-use.
clacky-ai/openclacky
Drive the desktop — macOS, or the Windows desktop through WSL — when a task needs a native app, dialog, menu bar, file picker or anything the terminal and browser tools cannot reach.
vellum-ai/vellum-assistant
Automate native macOS apps and system interactions via osascript (AppleScript)
Works with
Categories
Verify Reel, the macOS menu-bar screen recorder, by launching a disposable app bundle and driving its real UI with Computer Use. Verify Reel is an agent skill from rselbach/reel. Verify Reel, the macOS menu-bar screen recorder, by launching a disposable app bundle and driving its real UI with Computer Use.
Verify Reel fits situations like: tasks that involve Desktop control.
Run `npx skills add rselbach/reel --skill verify-reel -a claude-code`. Or copy the skill folder (.agents/skills/verify-reel in rselbach/reel) into .claude/skills/verify-reel in your project. Claude Code loads it when a task matches its description.
Run `npx skills add rselbach/reel --skill verify-reel -a codex`. Or copy the skill folder (.agents/skills/verify-reel in rselbach/reel) into .agents/skills/verify-reel in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add rselbach/reel --skill verify-reel -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/verify-reel, .gemini/skills/verify-reel, .github/skills/verify-reel and .opencode/skills/verify-reel in your project.
Going by SKILL.md and its folder, Verify Reel needs the command-line tools its instructions call (just).
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.
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. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
Verify Reel is published under the Unlicense licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 2k tokens (SKILL.md is roughly 8.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
Skills that share tags, products or a category with Verify Reel: Mac Computer Use (To3akaRin/mac-computer-use, 1.1k stars), Steer (disler/mac-mini-agent, 272 stars), Computer Use (kunpengtalk/PeakCode, 166 stars) and Openbridge Debug (AFK-surf/OpenBridge, 430 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
rselbach (a GitHub user) maintains it in rselbach/reel, which has 118 GitHub stars. The repository was last updated on September 3, 2026.
Source: rselbach/reel on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.