System

System

Open-source, MIT-licensed remote control that turns plain-English commands into real actions on your Mac.

65/100MonitorFreeFree

If you want your Mac to obey a Slack-length sentence from your phone, SYSTEM delivers that today, for free, with source you can read. The Computer Use Agent loop is what earns its keep — screenshot, analyze with Claude's vision, act, then verify against the real UI, rather than a brittle coordinate script. The core toolchain (shell, AppleScript, open_app, clipboard, screenshot, Apple Music, system controls) plus natural-language and cron scheduling covers the daily automation jobs most people actually run. Budget an evening for the three macOS permissions and setup; that's the toll for this much control. If you want a zero-config, managed assistant, SYSTEM is the wrong tool.

Verified 2d ago · liveness 65/100 · cite: rightaichoice.com/tools/system

Best for
  • Mac power users who want AI-driven desktop automation with full source access
  • Developers wiring Mac control into scripts and internal tools via the Bearer-token API
  • Remote workers who trigger automations from a phone or browser
  • Open-source fans who want a self-hosted, MIT-licensed alternative to closed assistants
Not ideal for
  • Windows or Linux users — SYSTEM is macOS only
  • Anyone who wants a cloud-only setup with no local Bridge install
  • Users who won't grant Accessibility, Screen Recording, and Automation permissions
Visit Website

IntermediateDesktop app path: the SYSTEM.dmg onboarding wizard walks you through the three macOS permissions and configuration, aiming at first value in a single short session. CLI path: clone the repo, npm install, npm run setup for the interactive wizard, npm start — expect an evening if you're new to the stack.Desktop · CLI · Web · APIAPI availableVerified 2d ago
Pricing
Free
FreeFree tier3 hidden costs
Learning curve
Intermediate
Desktop app path: the SYSTEM.dmg onboarding wizard walks you through the three macOS permissions and configuration, aiming at first value in a single short session. CLI path: clone the repo, npm install, npm run setup for the interactive wizard, npm start — expect an evening if you're new to the stack.
Runs on
DesktopCLIWebAPI
API available · 9 integrations
Who it's for
Remote worker with a Mac at homeDeveloper automating a Mac build/test boxPower user driving an app with no scripting interface
Live sentiment
Is System actually worth it?

We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.

  • Honest verdict, not marketing
  • Real pros & cons from real users
  • Attributed quotes with receipts
Run a free scan

3 free scans · no card needed

Skip it if

Skip SYSTEM if you need to control Windows or Linux machines, or want a cloud-only assistant with no local Bridge server and no macOS permission grants.

The 30-second take
Biggest gripe

You supply the Claude access behind the Agent's reasoning, so natural-language and computer-use calls bill to your own model account on top of the free MIT license.

Price reality

SYSTEM is MIT-licensed and free to run; the real cost is the Claude usage behind the Agent plus keeping a Mac powered on as the Bridge. For a solo Mac power user or developer wiring desktop control into scripts, that's cheaper than a paid closed assistant. Teams needing managed deployment or per-seat admin dashboards should look to hosted desktop-automation products instead.

In short

System — Open-source, MIT-licensed remote control that turns plain-English commands into real actions on your Mac. Best for Mac power users who want AI-driven desktop automation with full source access, Developers wiring Mac control into scripts and internal tools via the Bearer-token API, Remote workers who trigger automations from a phone or browser. Free to use.

What people actually say about System — is it worth it?

We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.

20 mentions across 6 sources (Hacker News, YouTube, Product Hunt, Stack Overflow, Lemmy, Tech Press), 81 more we could not attribute · researched Sep 30, 2026.

41% positive59% critical

Weighted by the 101 posts each of 6 sources contributed.

Recurring strengths
  • +MIT-licensed and completely free, with full source access and no vendor lock-in on the code
  • +Split cloud-reasoning / local-execution architecture keeps the LLM in the cloud and actions on your machine
  • +Computer Use loop with a screenshot-analyze-act-verify cycle is a genuinely capable automation primitive
  • +Accessibility-based targeting clicks elements by role or title instead of fragile raw coordinates
  • +Broad built-in toolset: shell, AppleScript, clipboard, notifications, Music, Calendar, Reminders, Notes, Finder
Recurring frustrations
  • −Zero substantive community reviews or GitHub discussion in the dataset to validate real-world behavior
  • −Requires granting Accessibility, Screen Recording, and Automation permissions — a big trust ask
  • −Setup depends on a bridge.config.json secret, a Cloudflare Tunnel, and a WebSocket bridge to get right
  • −macOS-only, so it cannot serve cross-platform teams or Windows/Linux workflows
  • −The Computer Use loop is inherently slow and token-hungry compared to a deterministic script
Patterns worth knowing
Most scraped posts are coincidental keyword matches, not discussion of this tool
Seen on Hacker News, YouTube, Product Hunt, Lemmy, Tech Press
Agents that touch real systems need safeguards before they get broad permissions
Seen on Tech Press, Hacker News
Local-first and self-hosted control is valued as autonomy rather than mere privacy
Seen on Hacker News, Lemmy
Learning curve
intermediateProductive in ~A few hours
Hidden costs people mention
  • • Your own Anthropic Claude API usage for the cloud agent, which scales with Computer Use screenshots and tokens
  • • Cloudflare Workers and Durable Objects usage beyond the free tier
  • • The time cost of the initial permissions, tunnel, and secret setup

Viability Score

65/100
Monitor

How well maintained and how widely used is System? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this

Recent activity
not measured
Traction
100
Site health
95
User sentiment
40
What the vendor publishes
20

Last calculated: October 2026

How we score →

Key Features

  • Natural-language control of macOS from a phone or browser
  • Computer Use Agent: screenshot → analyze → act → verify loop
  • Mouse automation: click, double-click, right-click, drag, scroll
  • Keyboard input via cua_type and key combos via cua_key (cmd+c, return)
  • Accessibility tools that find and click UI elements by role or title
  • Window management: move, resize, minimize, maximize, close
  • Menu item clicking by path (e.g., File > Save)
  • Schedule one-time and recurring tasks via natural language or cron
  • Persistent agent state, memory, and conversation history
  • Apple Music playback and volume control
  • System controls: brightness, dark mode, DND, lock, sleep
  • Calendar, reminders, notes, and Finder file operations
  • Shell command execution and AppleScript support
  • Text-to-speech, notifications, and clipboard get/set
  • WebSocket real-time bridge with Bearer-token API auth

About System

FreeIntermediateAPI availableDesktop · CLI · Web · API

SYSTEM is an open-source (MIT) remote control for macOS that turns natural-language instructions into real actions on your machine. You send commands from your phone or a browser; the reasoning happens in the cloud, and the clicking, typing, and scripting happens locally. The architecture is split deliberately: the Agent — the brain — runs on Cloudflare Workers, using Claude for natural language, with persistent state, memory, and scheduling handled by Durable Objects, reached over a WebSocket bridge. The Bridge — the body — is a local Express server on your Mac that executes AppleScript, shell commands, and Raycast extensions, exposed through a Cloudflare Tunnel. Every request is authenticated with an API secret sent as a Bearer token or a query parameter, generated during setup and stored in bridge.config.json. Two toolchains give you control. Core tools cover open_app, open_url, shell, applescript, notify, say, clipboard_get/set, and screenshot, alongside Apple Music playback, calendar, reminders, notes, Finder file operations, and system controls like brightness, dark mode, Do Not Disturb, lock, and sleep. The Computer Use Agent goes further: it takes a screenshot, analyzes it with Claude's vision, then clicks, double-clicks, right-clicks, drags, scrolls, types, and presses key combos in a screenshot → analyze → act → verify loop, with accessibility tools that target UI elements by role or title rather than raw coordinates. Setup runs through a menu-bar desktop app (SYSTEM.dmg, the recommended path) that walks you through the three required macOS permissions — Accessibility, Screen Recording, and Automation — or through a CLI wizard (git clone, npm install, npm run setup, npm start). It's macOS-only and MIT-licensed, aimed at Mac power users, developers, and remote workers who want scriptable desktop control and full source access rather than a closed assistant.

Behind the Verdict

SYSTEM is best understood as two halves that you own. The Agent half is a Cloudflare Workers service using Claude for natural language, with Durable Objects holding state, conversation history, and schedules, and a WebSocket bridge for real-time traffic. The Bridge half is a local Express server on your Mac that does the actual work through AppleScript, shell execution, and Raycast extensions, surfaced through a Cloudflare Tunnel. That split is the security story: your machine isn't exposed, and every request carries an API secret via Bearer token or query parameter. Strengths: the surface area is unusually wide for a free, MIT-licensed tool. Core tools include open_app, open_url, shell, applescript, notify, say, clipboard_get/set, and screenshot, plus Apple Music playback, calendar, reminders, notes, Finder operations, and system toggles for brightness, dark mode, DND, lock, and sleep. Scheduling accepts both natural language ("Every day at 5pm, play Closing Time") and cron syntax, and state persists across sessions — the /state endpoint even exposes preferences and history length. The Computer Use Agent is the differentiator: cua_screenshot, cua_screen_info, cua_get_windows, and cua_mouse_position for perception; cua_click, cua_double_click, cua_right_click, cua_drag, and cua_scroll for mouse work; cua_type and cua_key for keyboard; cua_focus_app, cua_launch_app, cua_window_manage, and cua_menu_click for app and window control; and accessibility tools that find elements by role or title instead of coordinates. Weaknesses: SYSTEM is macOS-only, needs Accessibility, Screen Recording, and Automation permissions granted by hand, and depends on your Mac being on and reachable while the Agent sits in the cloud. The CLI path assumes comfort with git, npm, and a terminal. Where it fits: Mac power users, developers embedding desktop control into scripts or internal tools via the Bearer-token API, and remote workers who want to trigger real actions from a browser or phone. Where it doesn't: Windows or Linux teams, anyone wanting a cloud-only setup with no local Bridge install, and teams expecting managed deployment or a fully guided, zero-config experience. The trade is hands-on configuration for full hackability at no cost.

Researching System? Get your full AI stack in 60 seconds.

Free, no signup — tell us your goal and get tools matched to your budget & existing stack.

Real-world workflow fit

Concrete scenarios for the personas System actually fits — and what changes day-one when you adopt it.

Remote worker with a Mac at home

From their phone browser, they send "Every day at 5pm, play Closing Time" and "Remind me to call mom in 30 minutes." The Agent on Cloudflare Workers parses the request, holds the schedule in Durable Objects, and fires the Bridge over the tunnel at the right time.

Outcome: Recurring Mac tasks run without the user touching the machine, and reminders surface on schedule.

Developer automating a Mac build/test box

They call the Bearer-token API to run shell commands and AppleScript over the WebSocket bridge, and schedule jobs with cron syntax via /schedules.

Outcome: Mac-side build and script tasks trigger remotely from scripts or internal tools, with state and history available through /state.

Power user driving an app with no scripting interface

They invoke the Computer Use Agent: cua_screenshot captures the window, Claude's vision analyzes it, then cua_click, cua_type, and cua_key act, and accessibility tools target elements by role or title.

Outcome: The task completes even in apps without AppleScript support, verified against the real UI rather than fixed coordinates.

Use Cases

Models Under the Hood

Claude

as of 2026-10-10

Limitations

  • SYSTEM is macOS-only and requires three macOS permissions — Accessibility, Screen Recording, and Automation — that you grant manually (the desktop app opens the right Settings pane for you).
  • It uses a split architecture: the Agent runs on Cloudflare Workers while the Bridge runs locally on your Mac, so your Mac must be on and network-reachable for commands to execute.
  • Setup is via the SYSTEM.dmg desktop app or a CLI path (git clone, npm install, npm run setup, npm start).
  • All requests need the API secret from bridge.config.json as a Bearer token or query parameter.
  • The CLI route assumes comfort with git, npm, and a terminal.

as of 2026-09-24

Verification history

We have re-verified System 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.

  1. — re-checked, vendor evidence unchanged
  2. — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
  3. — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
  4. — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
  5. — re-checked, vendor evidence unchanged
  6. — re-checked, vendor evidence unchanged

Showing the 6 most recent of 9 verification passes.

Free to cite with attribution — this page re-verifies continuously.

12-month cost

Project the real annual outlay, including the implied monthly cost when only an annual tier is published.

Annual total
Free
Over 12 months
Effective monthly
—
—

Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.

Plans compared

For each published System tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.

Free (open source, MIT)

$0

Ideal for

Mac power users, developers, and remote workers who want full source access and don't mind self-hosting the Agent and Bridge.

What this tier adds

Starting (and only) tier: the complete Agent and Bridge stack, Claude-powered natural language, the Computer Use Agent, scheduling, and API access — all at $0 under MIT.

Hidden costs & gotchas

What the public pricing page doesn't put in bold. Captured from pricing-page footnotes, contract terms, and recurring complaints.

  • You supply the Claude access behind the Agent's reasoning, so natural-language and computer-use calls bill to your own model account on top of the free MIT license.
  • The Bridge runs locally, so your Mac has to stay powered on and network-reachable — the machine itself is the ongoing cost of 'remote' control.
  • Computer Use runs a screenshot → analyze → act → verify loop, which means repeated vision calls per task and correspondingly more model spend than a single text command.

Where the pricing makes sense

The company stage and team size where System's pricing actually pencils out — and where peers do it cheaper.

SYSTEM is MIT-licensed and free to run; the real cost is the Claude usage behind the Agent plus keeping a Mac powered on as the Bridge. For a solo Mac power user or developer wiring desktop control into scripts, that's cheaper than a paid closed assistant. Teams needing managed deployment or per-seat admin dashboards should look to hosted desktop-automation products instead.

Setup time & first value

How long it actually takes to get something useful out of System — broken out by persona, not the marketing-page minute.

Desktop app path: the SYSTEM.dmg onboarding wizard walks you through the three macOS permissions and configuration, aiming at first value in a single short session. CLI path: clone the repo, npm install, npm run setup for the interactive wizard, npm start — expect an evening if you're new to the stack.

Switching to or from System

How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.

Migrating in
  • →From Apple Shortcuts: rebuild Mac automations as natural-language commands or scheduled tasks, and use shell/AppleScript tools where Shortcuts actions feel limiting.
  • →From Raycast scripts: keep Raycast, since SYSTEM's Bridge executes Raycast extensions, and add voice/browser control and scheduling on top.
  • →From custom AppleScript rigs: move trigger logic to the Agent and let the Bridge run the same AppleScript you already wrote.
Migrating out
  • ↗To Apple Shortcuts: port simple schedules and app launches back into Shortcuts if you no longer need remote or natural-language triggering.
  • ↗To a hosted desktop-automation service: replace the Cloudflare Worker Agent with a managed control plane, accepting less source-level control.
  • ↗To plain SSH plus scripts: keep the shell and AppleScript pieces and drop the natural-language and Computer Use layers.

Integrations

Apple MusicFinderNotesSafariCalendarRemindersMailRaycastLinear

Resources & Guides

Tutorials & Learning

YouTube returned 6 videos for “System”, and we withheld 6: 6 could not be judged, because “System” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about System.

Featured Head-to-Head Comparisons

Popular in Automation & Agents

Genspark

Genspark

Genspark is an all-in-one AI workspace turning one prompt into cited research, decks, dashboards and websites.

FreemiumTry
Air AI

Air AI

Air's Enterprise Readiness platform turns defense readiness data into a live Readiness Graph instead of slide decks.

Contact SalesTry
Cryptohopper

Cryptohopper

Cryptohopper runs an automated crypto trading bot in the cloud, with Copy Bot mirroring and an MCP server for AI trading agents.

PaidTry

Frequently Asked Questions

Used System? Help shape our editorial sentiment research.