Everything you can do with SuperPowers, in one clean guide.
This page documents the current feature set across the main web app, the iPhone / iPad app, the Apple Vision Pro app, the Android private streamer, the macOS streamer, and the desktop CLI streamer install flow. It is written for real users first: what each app does, how to set it up, what to say, which permissions matter, and how the apps fit together in daily use. This guide is kept in sync with shipped feature changes, including new marketplace, payments, and API surfaces.
What SuperPowers is
SuperPowers is a voice-first automation system that combines realtime chat, visual context, browser and device control, reusable prompts called powers, and private device streaming. Depending on which client you use, Super can:
See what you see
Web, iOS, and Vision Pro periodically summarize the live camera or panel view so the assistant can answer questions about what is visible.
Control browsers and devices
Use the web app directly, open browser agents on Apple platforms, or control connected Android and Mac private devices over WebRTC.
Package workflows as powers
Turn useful prompts plus voice commands into reusable powers, publish them to the marketplace, or save them privately in your prompt library.
Best starting path by platform
Use only the web app
-
1Open the dashboardGo to
app.getsupers.comand sign in with your phone or email. -
2Create or choose a projectProjects live in the sidebar and keep chats, powers, and device actions organized.
-
3Talk or typeUse the mic for realtime voice or the message box for chat. On desktop,
Entersends andShift + Entermakes a newline.
Add your own device
-
1Get a device slotUse your plan or unlock more private-device capacity if your account needs additional slots.
-
2Choose a streamer pathUse the desktop CLI from npm, the Android private streamer, or the Mac streamer depending on the device you want to host.
-
3Control it from chat or by tapping the streamOnce connected, Super can click, type, scroll, scrape text, take screenshots, and run supported device commands.
npm install -g superpowers-ai, then run superpowers login. The CLI signs you in or creates your account with your email or phone, starts the streamer, and prints a control link for /general.
Recent customer-facing changes
April 8, 2026
- Text your devices: if your account has text control enabled, you can send instructions like
open twitter in device 2orrun the marketing play in all devices. - Four-device grid by default: the remote-control area now opens with a four-device grid unless you choose a different layout.
April 7, 2026
- Desktop streamer from npm: install with
npm install -g superpowers-ai, then runsuperpowers login. - Instant control link: after login, the CLI starts the streamer and prints a browser link that opens the control surface directly.
April 6, 2026
- Chat attachments: use the
+button beside the message box to attach images, PDFs, and text files. - Keyboard control: when a remote Android, Mac, Windows, or Linux stream is active, browser keyboard input can be sent straight to that device.
- Theme switch: choose light mode or dark mode from Settings and the app remembers your preference.
April 5, 2026
- Chrome extension: a browser extension is now available as another setup option.
- What’s New bell: the web app now includes a notifications button for recent customer updates.
Which app does what
| Feature | Web | iPhone / iPad | Vision Pro | Android Private Streamer | macOS Streamer |
|---|---|---|---|---|---|
| Projects + chat | Yes | Yes | Yes | Embedded control panel mode | Yes |
| Powers + marketplace | Yes | Yes | Yes | Via embedded control panel | Voice prompt sheet + project chat; marketplace is centered on web / mobile clients |
| Integrations modal | Yes | Yes | Marketplace / studio integration path plus voice-connected Composio tools | Via embedded control panel | Yes |
| Camera or panel context summaries | Yes, every 10 seconds while eligible | Yes, every 10 seconds while eligible | Yes, every 10 seconds while eligible | No local assistant camera loop; it streams the device | No local camera loop; it streams the desktop |
| Private device hosting | Yes | Yes | Yes | Yes | Yes |
| Browser / panel agents | Browser-like power surfaces | Yes | Yes, in spatial panels | Embedded web control panel mode | No dedicated browser-agent layer |
| Tap-to-control remote stream | Yes | Yes | Yes | N/A | Yes |
| Fullscreen WebRTC | Yes | Yes, with double-tap | Spatial focus workflow | N/A | Main panel only |
| x402 paid APIs | Yes, full setup plus live device API listing | No dedicated native setup flow | No dedicated native setup flow | No | No |
How to use app.getsupers.com
The web app is the main command center. It gives you the most complete blend of projects, chat, powers, marketplace access, integrations, private-device control, camera context, and on-screen WebRTC controls.
Main navigation
- Projects sidebar: create projects, switch contexts, invite others, and manage account/project context without unrelated settings crowding the list.
- Top-right controls: model picker, powers, private-device control, and a new Settings gear live above the stream area.
- Chat composer: on desktop,
Entersends,Shift + Enterinserts a newline, and the+button lets you attach images, PDFs, and text files. - Mute: the mute button blocks mic input and also stops outbound realtime requests from being sent while muted.
- Settings + updates: the Settings modal now includes light mode and dark mode, and the bell icon opens recent customer updates.
Camera and visual context
- The browser can show a live camera preview.
- Every 10 seconds, when realtime is enabled and not muted, the app captures the current camera or active stream image.
- That image is summarized with
nvidia/nemotron-nano-12b-v2-vl:freethrough/tools/vision. - The text summary is silently fed into realtime so the assistant understands what is on screen without automatically speaking back.
Power Studio
Create and save powers
- Give the power a name.
- Add a voice command such as
post to twitter. - Write the prompt describing what the power should do.
- Optionally mark it for a remote device flow.
- Optionally enable x402 if you want to charge per run using USDC on Base.
- Use Publish to marketplace if you want other users to see and install it.
Import Claude Code / OpenClaw skills
- The Import Claude Code/OpenClaw Skills button accepts
.txt,.md, and.markdownfiles. - You can pick individual files or, where supported, an entire directory.
- Power Studio also accepts a direct
skills.mdURL. Paste a raw file URL, or a GitHub.../blob/...URL that points to the file, then click Load skills.md URL. - Each URL import fetches the file fresh and replaces the previous import from that same URL, so clicking it again updates the saved skills instead of leaving stale duplicates behind.
- The app combines all imported text files into one saved prompt with a source-file header.
- Large imports are trimmed automatically so the saved prompt stays usable.
Marketplace
Install and track powers
- The marketplace shows My Powers and All Powers.
- The web marketplace now opens full-screen, with denser Live Device APIs and My Powers cards so you can scan more items at once.
- The old inline x402 setup section is gone. Use the small Wallet + x402 button next to Refresh to open the dedicated setup modal instead.
- Each public power displays an Opens count.
- Installed powers show a visible checkmark.
- When you click Open, the power's prompt and voice command are also saved locally to your prompt list.
- On mobile web, the Powers entry point now stays reachable in both portrait and landscape, including Chrome on iPhone.
Creator workflows
- Published powers can be reopened, edited, unpublished, or republished.
- Marketplace creator rewards are surfaced in
PHONECLAWon the marketplace screen. - The marketplace now also shows x402 readiness, live device API inventory, and monetized power pricing where available.
- The Power Studio and marketplace are linked, so you can jump directly from an empty state into creating a new power.
Paid powers on web
Creator setup
- Connect a Base-compatible payout wallet.
- Set a price for the device API or power.
- Choose which public powers the live device is allowed to run.
- Keep the web host and private device connected while you want paid runs to work.
What buyers get
- Clear pricing in Base USDC.
- A payment challenge when payment is required.
- Execution on the creator's live connected device after payment succeeds.
- Optional directory visibility for public paid powers.
WebRTC stream controls
Record
Use the red record control to capture the active WebRTC stream. When you stop, the recording is analyzed with Gemini 2.5 via OpenRouter and converted into editable step-by-step instructions.
Fullscreen
Open the active stream fullscreen without covering it with large bottom controls. Mobile landscape keeps the stream in WebRTC mode and rotates the layout appropriately.
Close active
Close the active stream quickly from the top control row. Connected sessions remain managed through the private-device room list.
Keyboard control
When a remote device is active, browser key presses can be forwarded directly to Android, Mac, Windows, and Linux streamers.
Four-device grid
The remote-control area now opens with a four-device grid by default, so multiple live devices are easier to manage immediately.
Text-message control
If enabled for your account, you can send supported text commands and have them forwarded to your live devices through the same control stack.
openrouter/free for chat and control. Browser text-to-speech can read responses aloud, and the mic is briefly muted during speech to avoid feedback loops.
Support, payments, and private devices
- Support modal: opened from the sidebar, with a Calendly embed and quick links like YouTube, Discord, and the full docs at
app.getsupers.com/docs. - Connect Private Device: shows your purchased device count and room code, plus purchase actions for PayPal and crypto.
- Crypto device unlock: the crypto purchase path includes wallet connection and
PHONECLAWbalance checking. - iPhone visitors: the page can detect iPhones and try to redirect users into the App Store / iPhone app flow automatically.
The native Apple mobile app
The iOS client is the main native mobile experience. It combines AR camera context, chat, browser-agent surfaces, powers, integrations, private-device control, and a native onboarding flow backed by Superwall.
Onboarding and account flow
- The app includes a native onboarding sequence and routes users through the Superwall paywall during onboarding.
- Sign in with phone or email.
- If you use an email, the app can inspect the domain, infer the business, and present a confirmation sheet.
- After business detection, the assistant can greet the user personally and describe the inferred business.
Core UI
- Projects live in a sidebar with search and creation tools.
- Header actions include Marketplace, Integrations, Power Studio, stream controls, credits, add credits, invite, and private-device entry points.
- The app supports browser-style power windows and fullscreen web views in addition to camera-based AR overlays.
Voice, visual context, and mute behavior
- Realtime voice is available directly in the app.
- The app periodically summarizes the camera feed every 10 seconds using the same
nvidia/nemotron-nano-12b-v2-vl:freeimage model path used by the web app. - The assistant receives those summaries silently so it has scene context without forcing a reply.
- Mute parity is built in: muting blocks outgoing voice requests, output, and forced mic capture.
- When realtime is paused or blocked, the app can speak assistant replies through native text-to-speech while suppressing the mic to avoid hearing itself.
Private devices and WebRTC on iOS
- You can create private-device rooms, reuse saved room IDs, and connect Android or Mac streamers.
- WebRTC overlays support double-tap fullscreen.
- When fullscreen is active and the device rotates, the stream stays in WebRTC mode and rotates with the interface.
- Marketplace powers opened on iOS also get added to your local saved prompts, and installed powers show a checkmark.
AR and gesture features
- The iOS app includes AR-driven hand-pose and open-palm detection for its floating power / command surfaces.
- Browser windows, browser-status planes, and power surfaces can live inside the AR scene alongside camera context.
- This makes the iPhone / iPad app feel closer to a spatial control panel than a plain chat app.
The spatial version of SuperPowers
The Vision Pro app carries the same core system into a spatial panel with immersive passthrough, browser-powered workflows, power modals, voice tools, and panel-context summaries.
What you see in the main panel
- A web-powered main panel inside a rounded spatial surface.
- Quick buttons for Marketplace, Studio, Commands, Mute / Unmute, and Prompt.
- Optional overlays like maps and the powers menu.
- Onboarding and signup overlays if the session is not set up yet.
How voice works
- Voice prompt text is stored locally and can be appended into the assistant prompt.
- Mute blocks requests, output, and mic forwarding.
- Fallback speech uses native AVSpeechSynthesizer when needed.
- Voice commands can open marketplace, open Power Studio, control browser windows, and interact with private-device streams.
Panel context and browser agents
- Every 10 seconds, while voice is available and not muted, the current panel view is captured and summarized.
- The summary is sent into voice context as a silent
Camera update, even though in Vision Pro it is really a panel or browser view. - The app supports browser windows, focusing windows, scrolling, clicking, typing, opening fullscreen browser sessions, and returning them to 3D space.
- Marketplace and Power Studio are surfaced as modal flows controlled from the spatial panel.
Install the streamer from npm
The desktop CLI is the fastest way to turn your own Mac, Windows, or Linux machine into a private streamer without walking through a separate app UI first.
Install and sign in
-
1Install the packageRun
npm install -g superpowers-ai. -
2Log in or create your accountRun
superpowers loginand enter your email or phone plus the verification code. -
3Open the printed control linkThe CLI starts the streamer and prints a link that opens the control surface directly in the browser.
Useful commands
npm install -g superpowers-ai superpowers login superpowers start superpowers stop superpowers whoami superpowers logout
Your private Android device host
The Android private streamer is the Android app you install on your own phone to let Super stream and control it remotely over WebRTC. It can also switch into an embedded client mode that loads the main control panel for the same account.
Host mode screen
- Account field for phone or email.
- Send code and Verify sign-in flow.
- Status rows for signaling, room, streaming, and Moondream.
- Enable Accessibility, Disconnect, and Open Control Panel buttons.
- A device-unlock purchase button appears when the account has no available device slot.
Client mode
- The app can switch from streamer mode into a WebView control panel.
- That control panel loads the main web app with the current owner ID prefilled.
- Back to Streamer returns to host mode.
- Reload refreshes the embedded control UI.
Permissions and reconnect behavior
- Accessibility is required for text entry, tapping, long pressing, multi-clicking, scrolling, and scraping UI text.
- Screen recording is required for WebRTC streaming.
- Microphone may be requested for audio capture.
- Notifications can be requested where needed by the app flow.
- After asking for screen-recording permission, the app waits 10 seconds and tries to click
Start nowautomatically through the accessibility layer. - The app runs an auto-rejoin loop every 15 seconds so it can reconnect when the host refreshes or a room temporarily disappears.
Remote commands the Android streamer supports
Navigation
Open app, open URL, Home, Back, and scrolling.
Touch actions
Single tap, 5-second hold / long press, and rapid multi-click bursts such as 10 to 20 taps.
Text and extraction
Type text, send text, scrape targeted text, scrape page text, and capture screenshots.
Your private Mac desktop host
The macOS streamer is the desktop companion for streaming and controlling your own Mac. It includes projects, chat, a model picker, Composio integrations, voice command settings, and a WebRTC panel that can be remote-controlled by tapping the live stream.
Main app layout
- Sidebar: projects, credits, device counts, unlock-devices button, logout.
- Top row: model picker, voice toggle, Commands, Integrations, Private Device, Cloud Agent / Stop Stream.
- Chat panel: native project chat with message history and send box.
- WebRTC panel: shows the remote stream and accepts tap-to-click on the live video.
Private-device flow
- Use Private Device to display the room code for pairing.
- The sheet includes the Android streamer download and a tutorial link.
- Cloud Agent starts hosting the room if you have a device slot.
- The app now includes auto-reconnect logic so it can recover when the remote host refreshes or the signaling room drops.
What the Mac streamer can do remotely
| Capability | What it means in practice |
|---|---|
| Open apps and URLs | Launch installed apps by name or bundle, or open web links. |
| System-wide input | Click, long press, multi-click, scroll, type text, and send keyboard shortcuts through accessibility and CGEvent APIs. |
| Terminal hooks | Open Terminal, focus it, type directly into it, and optionally press Enter. |
| Perception helpers | Take screenshots, run object-targeted click detection, and scrape visible UI text from the frontmost app. |
How powers work from creation to install
Create a power
- Add a title that you can recognize later.
- Add a voice command if you want to trigger it by speech.
- Write the prompt in plain language.
- Choose whether it is mainly for remote-device workflows.
- Choose whether to expose it through x402 and set a Base USDC price per run.
- Choose whether to publish it to the marketplace for everyone.
Install a power
- Open the marketplace and click or tap Open.
- The power is marked as installed with a checkmark.
- The marketplace open count increments.
- The prompt and voice command are copied into your local saved prompts so you can reuse or edit them later.
Enable a live device API
- Click Use wallet to connect a Base-compatible
0x...payout wallet for the creator account. - Turn on x402 for the user profile in the marketplace status area using the checkbox. The toggle now persists immediately.
- Set the device's Base USDC price, provide a Base payout wallet, click Save x402, and keep the web controller host connected to a live WebRTC room.
- You can save the Base wallet and price before the enable toggle is turned on.
- Use the marketplace cards to select which public powers your API will accept.
- Once a device is attached, the room appears in the marketplace as a live API target.
What x402 adds
- Public powers can be sold per run in Base USDC.
- Device APIs and powers become callable like paid endpoints.
- Each x402-enabled user now has a supported-powers allowlist that controls what their live device API will run.
- The backend returns an HTTP
402challenge when payment is missing or invalid. - After settlement, the run is pushed directly to the live device host over WebRTC.
Stream-to-steps workflow on web
-
1Record a live streamUse the WebRTC record button in the web app while the target task is happening.
-
2Let Gemini 2.5 analyze itThe recording is sent through OpenRouter using Gemini 2.5 to produce step-by-step instructions.
-
3Edit and saveThe generated steps appear in an editable modal. Click Add to Saved Prompts when it looks right.
Paid device API flow
-
1Connect a payout walletAdd a Base-compatible wallet and set your device API price from the marketplace or Power Studio.
-
2Enable the powerChoose which public powers your live device is allowed to run for paid requests.
-
3Keep the host liveYour web host and private device need to stay connected so the paid run has a live target.
-
4Share the paid APIOnce payments are enabled, buyers can trigger the power against your live device using the published paid API.
What creators need
- A valid Base payout wallet.
- A price set for the live device API or power.
- A live web host with an attached private device.
- The supported paid powers enabled for that device.
What buyers see
- The creator name and price before payment.
- The device or power being requested.
- A normal payment challenge if payment is required.
- The result after the live device finishes the run.
When I say post launch update, open X, draft a short update about today's release, include one clear CTA, and post it.
How speech, fallback, and mute work
Normal realtime behavior
- Voice can run in realtime where the client supports it.
- Scene summaries from camera or panel context are silently added so Super understands what you are looking at.
- Business welcome prompts can be spoken automatically after login and account confirmation.
- Realtime metering auto-recovers after a browser reconnect or backend restart instead of surfacing a stuck meter-session error.
Fallback behavior
- When realtime credits are blocked, the web app can fall back to
openrouter/free. - Web uses browser text-to-speech for spoken fallback.
- iOS and Vision Pro use native speech synthesis for fallback.
Useful things to say
General
Open Marketplace Create a power called invoice follow-up Open profile Add credits
Device control
Focus device 1 full screen Open Terminal and type npm run dev Long press the submit button for 5 seconds Click the follow button 15 times
How personal device streaming works
Private devices let you connect your own Android phones or Macs to Super. This is separate from cloud devices and is designed for personal hardware you control.
-
1Get a device slotYour account needs an available device slot. This can come from direct purchase or the eligible crypto path where supported.
-
2Generate a roomFrom web, iOS, or Vision Pro, open the private-device flow and generate or reuse a room code.
-
3Pair the streamerEnter that room on the Android private streamer, start hosting from the Mac streamer, or use the desktop CLI for Mac, Windows, or Linux.
-
4Control the streamUse chat, voice, text-message commands where enabled, keyboard input, or direct taps on the WebRTC view. On web, top controls also let you record, fullscreen, or close the active stream.
What happens on disconnect
Both Android private streamer and Mac streamer now focus on reconnecting to the same room instead of dropping permanently when the host refreshes or the signaling room temporarily disappears.
What you can do from the stream view
Tap or click the stream to send normalized pointer actions. The assistant can also focus a device by number, open it fullscreen where supported, and dispatch device-specific commands over the data channel.
Connecting external services
Integrations are handled through Composio. The product shows recognizable service tiles with favicons and green checkmarks when a service is already connected.
Common built-in tiles
Gmail, Google Drive, Google Calendar, Slack, Notion, GitHub, Linear, and Jira are surfaced prominently in the main UIs.
How connection works
Open the Integrations modal, tap the service, and the app opens the login or connection flow. Once complete, the tile shows as connected.
Why it matters
Connected services become available to powers, scheduled commands, and assistant workflows so Super can act across your tools.
How credits and device access are unlocked
Credits
- The apps surface credit status and can direct you to add credits through profile or payment flows.
- After payment, credits may take a short period to appear.
- If realtime credits are unavailable, free model fallback may still let you chat and control devices, depending on the client and model selection.
Device slots
- Private devices consume device slots.
- Purchase links are surfaced in the app when no slot is available.
- On web, the Connect Private Device modal includes a PayPal purchase option and a crypto purchase path.
- The current web lifetime-deal countdown is set to end on March 23, 2026 at 12:00 AM in the visitor's local browser time.
Wallet and PHONECLAW details
- The web app supports connecting Phantom and OKX Solana extension wallets.
- Wallet flows can verify whether the connected wallet holds the required value of
PHONECLAW. - When eligible, the wallet path is used to unlock device access in the private-device flow.
- Wallet connection and verification are separate from the older support and lifetime-deal surfaces; the current private-device purchase flow is the main place where this matters for users.
- These Solana wallets are for private-device unlocks and credit additions; x402 creator payouts now use a separate Base-compatible
0x...wallet.
x402 payment readiness
What you need
- x402 must be enabled on the server.
- The creator must save a valid Base-compatible
0x...payout wallet. - The creator must enable x402 and choose a device API price plus payment token.
- The power must be public if it is going to be sold through
/x402/run.
Status indicator meanings
- x402 enabled: payments are turned on for paid runs.
- Payments ready: the app has enough information to accept paid requests.
- Wallet linked: the creator has a payout wallet attached.
- Live devices: the count of connected device APIs currently available to paid callers.
- Supported powers: how many marketplace powers the current x402 device API is willing to run.
What each permission is for
| Permission | Where it matters | Why it is needed |
|---|---|---|
| Camera | Web, iOS | Live camera preview plus periodic scene summaries for assistant context. |
| Microphone | Web, iOS, Vision Pro, Android streamer | Voice input and, on some device streamers, audio capture. |
| Screen Recording | Android private streamer, macOS streamer | Required for WebRTC desktop or phone streaming and for screenshot capture on Mac. |
| Accessibility | Android private streamer, macOS streamer | Needed for taps, typing, long presses, multi-clicks, scraping text, and other UI automation. |
| Notifications | Android private streamer | Requested for app workflow support where needed. |
Fast answers to common issues
I am muted, but the assistant still seems active. Why?
On current web, iOS, and Vision Pro builds, mute is intended to block outgoing realtime requests as well as audio playback. If something still seems active, unmute once, then mute again to reapply the request block and mic suppression state.
Why does a private device say room not found?
The room may have been refreshed or dropped. Current Android private streamer and Mac streamer builds include an auto-reconnect loop to retry the room instead of failing permanently. Wait for the reconnect cycle or recreate the room from the host client.
Why can I not control my Android or Mac device?
Most control actions need Accessibility permission. Streaming itself also needs Screen Recording on Android and macOS. If those permissions are missing, taps, typing, screenshots, and text scraping will fail even if the room is connected.
Why is a marketplace power not showing as installed?
Open the power from the marketplace. Installed powers are tracked locally and should show a checkmark on web and iOS after the open action saves the prompt and voice command into your saved prompt list.
What happens if I run out of realtime credits?
Depending on the client, the app can fall back to free OpenRouter chat and local text-to-speech. The web app is designed to default toward the free path when paid realtime is unavailable, rather than blocking basic chat entirely.
Why is my paid power or device API not working?
Most paid-run issues come down to one of four things: the payout wallet is missing, the price was not saved, the power is not enabled for the device, or the web host and private device are not both live. Re-save the marketplace settings, make sure the device is attached, and try again.
How do I make a power available for paid runs?
Open the power in the marketplace or Power Studio, make it public, turn on payments, set the price, and mark it as supported for the live device that should run it.
Why did a paid request fail after payment?
That usually means the live host disconnected, the private device dropped, or the power was no longer available to that device. Reconnect the host and device, then retry the run.
Why is my device not showing up in Live Device APIs?
Today the x402 live-device list only includes active web-controller rooms. Make sure you enabled x402, saved a Base-compatible payout wallet, set a device API price, and kept the web host connected to a live WebRTC room with an attached device.
What if the x402 checkbox in the marketplace does not stay on?
If the toggle flips back off, the most common causes are a missing Base wallet, an unsaved price, or editing the power while signed into the wrong account. Save the wallet and price first, then enable payments again.
Why does x402 say the device API has not enabled that power?
The live x402 device API now uses an allowlist. Open the marketplace while signed in as the device owner, enable x402, and click the support toggle on the public power you want that device API to accept.
How do I get human help?
Open the Support modal from the web app sidebar. It includes a Calendly embed for booking help plus quick links to Discord, YouTube, and the docs at app.getsupers.com/docs.