Cheat Layer
Cheat Layer turns plain-English instructions into working browser and desktop automations. Its Project Atlas framework builds agents that scrape leads, publish social content, and run email, SMS and voice sales conversations across Windows, Mac and Linux.
What is Cheat Layer?
Cheat Layer is a no-code automation platform that converts an instruction written in ordinary English into an automation that actually runs. Its own framework, Project Atlas, was published in July 2022 and is presented as the first published agent framework, backed by an in-house model the company calls Atlas-1. The company also claims that in August 2021 it became the first startup approved by OpenAI to sell GPT-3 for automation. It was founded during the pandemic to help people who had lost jobs or businesses rebuild online, after the founders had assisted hundreds of them free of charge.
The documentation gives the product a second name, Open Agent Studio, described as the first cross-platform desktop application for Agentic Process Automation and positioned as an open-source alternative to UiPath. Its distinguishing idea is the semantic target: instead of the brittle code selectors that conventional RPA tools depend on, the intent of a target is expressed in English, so an agent keeps working after a site is redesigned. Around that sit semantic triggers, reasoning-based targets interpreted by a language model, a drag-and-drop node editor with loops and conditional logic, a Python node, a Magic Scraper, GPT-4 data processing, global variables and community agent sharing.
Three families of one-click cloud agents are offered. Marketing agents generate, post and A/B test content using services such as Leonardo AI, HeyGen, ElevenLabs, Pexels, DALL-E and Stable Video. Sales agents handle email, SMS, embedded chat with voice, and GPT-4 telephone calls through Twilio. Product agents use a Live Mode that iteratively builds tools, apps, presentations and landing pages hosted on the spot. Open Agent Cloud goes further, generating a no-code agent from a recorded Loom video and cloning a schedule across thousands of parallel agents in prepared Linux containers.
It reaches users through a Chrome extension opened with Alt+S on any page, desktop applications for Windows, Mac and Linux, a web dashboard and a JSON API. The extension counts around 8,000 users and rates 4.63 out of 5 across 57 reviews. The vendor claims more than 10,000 users, CPM 18 to 36 times better than TikTok ads and twice the reply rate of Apollo.io. One caveat matters: that extension now carries the name SuperPowers AI, and every call to action on the site leads to getsupers.com.
What it does
- Build an automation by describing the goal in ordinary English
- Click, type, open tabs and extract data on any website using semantic targets
- Scrape B2B leads and enrich them through several third-party APIs
- Generate, publish and A/B test social content daily across TikTok, Twitter, Facebook and YouTube
- Deploy autonomous email, SMS, chat and telephone sales agents
- Generate apps, websites, browser extensions and mobile apps from a conversation
- Schedule agents on UTC cron, trigger them by webhook, and run them in parallel in the cloud
When to use Cheat Layer / When not to
A quick filter to help you decide if Cheat Layer is the right fit.
When to use Cheat Layer
- Small businesses with no developer and no agency budget that still want repetitive work automated
- Growth marketers running daily A/B tests and automated publishing on TikTok, Twitter, Facebook and YouTube
- Outbound sales teams deploying email, SMS, chat and telephone agents that hold conversations unattended
- Content creators generating and repurposing video and SEO articles at volume
- RPA practitioners who want targeting that survives interface redesigns instead of brittle code selectors
When not to use Cheat Layer
- Regulated organisations needing documented GDPR compliance, a DPA or a subprocessor list, none of which exist here
- Buyers who require a published price, since the main plan shows no rate anywhere on the site
- Users unwilling to grant sweeping permissions such as debugger, screen capture and full disk access
- Teams that need a stable product identity, as the brand is being folded into SuperPowers AI
- Genuine non-technical users, despite the no-code promise, because Python, third-party API keys and UTC cron scheduling are required
How to use Cheat Layer
A typical end-to-end flow, from setup to results.
- Create an account on cheatlayer.com, accepting the licence agreement and privacy policy
- Install the Chrome extension from the Chrome Web Store
- Hold Alt+S, or Option+S on a Mac, to open Cheat Layer on any website
- For desktop use, download Open Agent Studio from GitHub and, on Windows, install Python 3.10 from the Microsoft Store
- On a Mac, grant Terminal the Accessibility, Screen Recording and Full Disk Access permissions, then allow the app under Privacy and Security
- Collect your API key at the bottom of the billing page, or set it as the CHEAT_KEY environment variable on Linux
- Describe the automation in plain English, or assemble it node by node in the visual editor
- As a shortcut, upload a Loom recording of the process and let Open Agent Cloud generate the agent
- Add the keys for whichever third-party services you use, such as Twilio, Supabase, HubSpot, Vercel, GitHub, OpenAI or ElevenLabs
- Schedule the agent on UTC cron or trigger it by webhook, then follow runs in the history and dashboard
Pros & Cons
Pros
- The semantic targeting approach genuinely addresses the fragility of conventional RPA selectors
- Unusually broad coverage for a vendor of this size: browser, desktop, cloud and API across three operating systems
- The Open Agent Studio desktop application is distributed openly on GitHub and presented as open-source
- A simple, documented JSON API that can be exposed locally through ngrok
- Generating an agent from a Loom recording makes for a very short path to a first result
- A generous and clearly quantified creator programme paying 30% recurring plus one dollar per thousand views
- The extension is actively maintained, with version 18.10.65 released on 7 August 2026 and no violation history
Cons
- The brand identity is confused: the site sells Cheat Layer, the documentation says Open Agent Studio, the extension is now SuperPowers AI and every button leads to getsupers.com
- The main plan carries no published price, the pricing link points nowhere and the signup page renders only in JavaScript
- Stated prices and live checkouts disagree: the email agent is advertised at $149 per month but billed at $199, and the phone agent is charged per day
- The licence agreement is a botched find-and-replace of GitHub's Atom agreement, leaving broken URLs, a malformed support address and stray references to Atom and the Octocat
- Two different corporate names appear in that same document, alongside three inconsistent jurisdictions and no postal address at all
- Compliance material is absent: no GDPR mention, no DPA, no subprocessor list, no certification, and a privacy policy frozen since January 2022
- The permission surface is as wide as a browser extension can request, and several advertised agents remain marked coming soon
Pricing & Plans
There is no confirmed permanent free plan: the published table contains no free tier, although one page still shows a Signup Free button. The cheapest paid entry point actually published is USD 9 per month for the TikTok marketing agent, a figure confirmed by the vendor's own checkout. The main Unlimited Workflow Automation Cloud plan carries no published rate, and Enterprise is quoted on request. All plans are advertised with a 60-day money-back guarantee, and no free trial is offered.
- Unlimited Workflow Automation Cloud — price not published. Unlimited agents and executions
- one Desktop Cloud Agent Server covering 200 cloud agents per day
- additional servers at USD 3 per day
- Agents API access
- unlimited webhook triggers
- unlimited Google Sheets tasks and data
- unlimited email
- SMS
- Enterprise — custom pricing. Everything in the workflow plan plus a bespoke SLA
- a private office-hours calendar
- custom team seats
- SSO SAML and the option to run Cheat Cloud on-premise
- TikTok Marketing Agent — USD 9 per month
- with daily automated A/B testing
- Twitter Marketing Agent — USD 9 per month
- announced as coming soon
- Cold outbound lead generation and sales agent — advertised at USD 149 per month for 10
- 000 emails and announced as coming soon
- although the live checkout charges USD 199 per month
- Desktop Cloud Instance — USD 90 per month at checkout
- matching the USD 3 per day quoted for an additional server
- Phone Agent — USD 449 per day at checkout
- a billing period that looks like a vendor configuration error
- White Label and Agency plans — referenced on the billing page as the way to remove Cheat Layer branding from embedded agents
Data, GDPR & hosting
A consolidated view of how Cheat Layer handles your data.
GDPR overview
There is no mention of the GDPR anywhere on this site. Searches across the privacy policy, the licence agreement and the documentation, in both extracted text and archived HTML, return no occurrence of the regulation, and none of the CCPA either. No legal basis for processing is stated, no right of access, rectification, erasure or portability is described, no Article 27 EU representative is designated and no data protection officer is named. No data processing agreement is offered, no subprocessor list is published, and international transfers are handled by a generic consent clause rather than standard contractual clauses. The policy is dated 14 January 2022 and has not been revised since. The Chrome Web Store listing compounds this by declaring the publisher a non-trader, warning that EU consumer rights do not apply.
Who owns the data?
The end-user licence agreement states that the software is licensed rather than sold and that the vendor reserves every right it does not expressly grant. Users may not resell, rent or exchange the software, may not resell accounts, and may not decompile or reverse-engineer it. Notably, no clause addresses ownership of the automations, agents or output that customers themselves create, which leaves that question open. The privacy policy allows personal data to be shared with service providers, with corporate affiliates including the parent company and subsidiaries, and with business partners, and to be transferred wholesale in a merger, sale, financing or bankruptcy. Open-source components remain governed by their own licences, which prevail over the agreement.
Reuse rights
The licence permits installing as many copies as wanted and using the software for any lawful purpose, and the desktop application may be redistributed provided the agreement travels with it. One tension deserves attention: the agreement forbids selling or exchanging the software for anything of value, while the marketing pages actively encourage users to publish premium agents for recurring revenue, to embed agents and sell access on their own websites, and to turn automations into Chrome extensions or desktop applications they could sell. Nothing in the documents explicitly grants or denies commercial reuse of agent output, so that permission rests on the marketing copy rather than on the contract. Operational limits do apply: webhooks are capped at one per minute, multiple accounts are prohibited, tenants must belong to the same business and share its domain, and installing locally or on-premise requires an enterprise licence. Data drawn through Google APIs is subject to Google's API Services User Data Policy and its Limited Use requirement.
Data retention & training
Hosting summary
The vendor declares no hosting country or region. The privacy policy says only that data is processed at the company's operating offices and in any other place where the parties involved are located, and that submitting information constitutes consent to that transfer. Network resolution tells us more than the documents do, and should be read as an observation rather than a vendor commitment: the domain resolves to 34.170.154.36, an address belonging to Google LLC in Council Bluffs, Iowa, which is a Google Cloud region rather than a content delivery node. Desktop installers are served from Google Cloud Storage. Open Agent Cloud runs agents in Linux containers on the vendor's servers, with no location stated. Where Supabase is used as a vector store for sales conversations, that data sits in the customer's own Supabase account. Enterprise customers may run Cheat Cloud on-premise for compliance reasons. Encryption at rest or in transit is never mentioned.
Things to keep in mind
Risks and trade-offs to weigh before adopting Cheat Layer.
- The browser extension requests about the widest access possible: the debugger permission exposes page content and network traffic through the developer protocol, desktopCapture allows screen recording, and host permissions cover every URL you visit
- On macOS the official setup asks you to grant Terminal accessibility, screen recording and full disk access, system-level concessions that most people never revisit once given
- The product automates third-party platforms and scrapes competitors' profiles, which collides with the terms of service of those services and can get accounts banned, a risk the site never raises
- Sending tens of thousands of automated emails and running telephone agents raises consent obligations that the site does not address anywhere
- No GDPR mention, no data processing agreement and no subprocessor list, while the homepage loads several undisclosed third-party trackers alongside an API key left in plain sight in public HTML
- Handing commercial and social accounts to agents that converse unsupervised invites dependence and a quiet erosion of the operator's own judgement about what is being published in their name
- Longevity is uncertain: the brand is being absorbed by a successor product, the original blog no longer responds, and several advertised features have been coming soon for a long time
Setup & Integrations
Technical difficulty
Two very different paths exist. Installing the Chrome extension and deploying a one-click cloud agent through a form is easy and works from a phone. The full path is considerably harder: Python 3.10 on Windows, an unsigned binary to unzip and force past the operating system, system-level permissions on macOS, an environment variable on Linux, an API key fetched from the billing page, keys for every third-party service used, a Supabase vector database for autonomous sales agents, and scheduling expressed in UTC cron.
Deployment
Integrations
Supported languages
Behind Cheat Layer
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Cheat Layer.
Frequently asked questions
What does Cheat Layer actually do?
Do I need to know how to code?
Where does it run?
Is there an API?
How much does it cost?
Is there a free trial?
What permissions does the browser extension request?
Is the tool GDPR compliant?
Is my data used to train models?
What is the relationship with SuperPowers AI?
Should you pick Cheat Layer?
Cheat Layer rests on a genuinely interesting technical idea. Conventional robotic process automation breaks whenever a website changes its markup, and describing the intent of a target in English rather than pinning it to a selector is a real answer to that problem. Around that idea the vendor has assembled an unusually wide product for its size: browser and desktop automation on three operating systems, one-click cloud agents for marketing, sales and product work, a documented JSON API, agent generation from a screen recording, and a desktop application published openly on GitHub.
The surrounding execution is much weaker, and the gap is hard to ignore. The licence agreement is a failed search-and-replace of another company's document, with broken links and a malformed contact address. Two different corporate names sit in that one file, three jurisdictions contradict each other, and no postal address is published anywhere. There is no mention of the GDPR, no data processing agreement and no subprocessor list, while the homepage quietly loads half a dozen third-party trackers. Pricing is opaque: the flagship plan shows no figure, and where figures exist the page and checkout disagree.
The permission surface deserves a clear-eyed look too. Such a tool needs deep access, but debugger rights on every URL plus screen capture, and full disk access granted to Terminal on macOS, is close to the maximum a user can concede. The product also automates third-party platforms in ways their terms of service tend to forbid, and that risk sits with the operator.
Most decisive is the trajectory. The extension now ships as SuperPowers AI and every button leads to another domain. Cheat Layer is best read today as a capable engine whose brand is being retired around it: worth a technically confident experimenter's time, difficult to recommend to a regulated organisation.
- Choosing a selection results in a full page refresh.
- Opens in a new window.