Can I use OpenCluely with any AI model?

Currently OpenCluely is built around Google Gemini (Flash-Lite and Flash models). Support for OpenAI, Anthropic Claude, and local models is planned.

Does it work in non-English languages?

Gemini understands most major languages, so you can ask questions in any language. Whisper voice transcription supports multiple languages with automatic detection.

How does OpenCluely compare to Natively, Interview Coder, FinalRound AI, and LockedIn AI?

OpenCluely is the free, MIT-licensed open-source option in the same category as Natively, Interview Coder, FinalRound AI, and LockedIn AI. It focuses on invisibility, voice + screenshot input, and real-time streamed Gemini answers. Unlike the paid tools there is no subscription, no account, and no telemetry. Unlike source-available clones, OpenCluely is genuinely MIT-licensed and accepts community contributions.

Does OpenCluely support BYOK (bring your own key)?

Yes. BYOK is the only mode. Paste your own Google Gemini API key into Settings (or the .env file) and OpenCluely uses it directly. There is no proxy server, no OpenCluely-side account, and no usage-based billing from us. Your key, your data.

Will it get me caught in an interview?

OpenCluely is intended for learning, practice, and authorized use. Always follow the rules of any interview you take - many employers and proctoring tools forbid external assistance, and modern proctoring (eye tracking, copy-paste detection, AI-text fingerprinting) can catch improper use regardless of how hidden the overlay is.

Does it work offline?

The local Whisper speech-to-text option runs entirely offline. The AI answers themselves need an internet connection because they call the Google Gemini API.

Can I customize the keyboard shortcuts?

Yes. Open the Settings window from the system tray or with Cmd/Ctrl+, to rebind screenshot capture, toggle voice, toggle window visibility, click-through, chat, and more.

Free and open source · MIT

The invisible AI interview copilot

OpenCluely is the free, open-source Cluely alternative. A stealth overlay that Zoom, Google Meet, Microsoft Teams, Discord, and OBS cannot see. Ask by voice or screenshot and get clear, streamed answers from Gemini in real time.

Download View source

Invisible to screen share
Gemini powered · BYOK
Windows, macOS, Linux
MIT · no telemetry

Cmd Shift S DSA C++

OpenCluely

"Find the duplicate number in an array of n plus one integers."

Floyd's cycle detection runs in O(n) time and O(1) space.

A real look at the overlay. Nothing here is captured by screen recording.

Stays hidden on

Zoom Google Meet Microsoft Teams Discord OBS Zoom Google Meet Microsoft Teams Discord OBS

Everything you need, nothing you do not

Built to be fast, quiet, and genuinely helpful under pressure.

Invisible overlay

Windows stay out of Zoom, Meet, Teams, Discord, and OBS captures. You see the answer, the call does not.

Real-time voice

Speech is split on natural pauses, not a blind timer, so a single spoken question stays a single question.

Streamed answers

Replies appear word by word as the model thinks, in both the chat and the floating answer window.

Direct image analysis

Screenshots go straight to Gemini for visual reasoning. No slow OCR step in the middle.

Session memory

It remembers the whole conversation, so follow-ups, edge cases, and optimizations all keep context.

Process disguise

Runs under ordinary system names and ships with no telemetry. Everything stays on your machine.

Three steps, then you forget it is there

Set it up once and let it sit quietly in the corner.

Ask

Speak the question or press the screenshot shortcut. The mic listens for natural pauses on its own.

Reason

Gemini reads the audio or image with full conversation context and works toward a precise answer.

Answer

The response streams into the overlay in real time, with clean code blocks and syntax highlighting.

Download the latest release

Pre-built installers for every platform. These links always point at the newest version.

Fetching the latest release from GitHub

See it stay hidden

A short capture of the overlay running live during a screen share.

Questions, answered

The things people usually want to know before trying it.

Is OpenCluely really invisible during screen sharing?

Yes. The overlay windows are excluded from capture in tools like Zoom, Google Meet, Microsoft Teams, Discord, and OBS. When a screen share starts, the app can also hide every window on its own. You still see the answers on your physical display.

Do I need an API key?

You need a free Google Gemini API key for the AI features. On first launch the app opens its Settings window and walks you through adding it. You can paste it in the UI or edit the .env file directly.

Is voice input required?

No. Voice is optional. You can use screenshots and the chat on their own. If you want speech, pick either local Whisper for offline transcription or Azure Speech. When no provider is set up, the mic button hides itself.

Which platforms are supported?

Windows and Linux (.deb and AppImage) ship as pre-built installers, built automatically on GitHub Actions with every release. macOS is not pre-built — the unsigned app is blocked by Gatekeeper, so on macOS you run it from source with a one-line ./setup.sh.

Is my data collected?

No telemetry and no data collection. Processing happens locally and your session stays on your device. Only the requests you send to the AI provider leave your machine, encrypted in transit.

Is it free?

Yes. OpenCluely is open source under the MIT license. You only pay for whatever AI usage your own Gemini key incurs, which has a generous free tier.

How does the screen-capture invisibility work?

On Windows OpenCluely uses WDA_EXCLUDEFROMCAPTURE and on macOS it uses NSWindowSharingNone — both are operating-system-level flags that tell the screen recording APIs to skip that window. The app checks for active screen sharing and can auto-hide every window when a share begins. Linux does not offer this API, so the overlay is visible on that platform; share a single window instead of the full screen as a partial workaround.

Is OpenCluely a free Cluely alternative?

Yes. OpenCluely is a free, open-source Cluely alternative under the MIT license. It does the same core job — a stealth overlay that captures your audio and screen, sends it to an LLM, and shows real-time answers — without the subscription, account, or telemetry. You bring your own Gemini API key (BYOK), which keeps the cost near zero on the free tier.

Is it really undetectable on Zoom, Meet, and Teams?

On Windows the overlay uses the WDA_EXCLUDEFROMCAPTURE flag (via SetWindowDisplayAffinity) and on macOS it uses NSWindowSharingNone — both are OS-level APIs that exclude the window from screen capture. Zoom, Google Meet, Microsoft Teams, Discord, and OBS skip those windows. The app also auto-hides every window when a screen share begins. Linux lacks this API so the overlay is visible on that platform.

Does it work with HackerRank, LeetCode, and CodeSignal?

The screenshot capture works in any application — HackerRank, LeetCode, CodeSignal, CoderPad, Codeforces, your own editor, a PDF, a slide. Anything you can see on screen, OpenCluely can capture and send to Gemini for analysis.

Prefer to build it yourself?

Three commands and the setup script handles the rest.

Clone the repository

git clone https://github.com/TechyCSR/OpenCluely.git
cd OpenCluely

2
Run the setup script
```
./setup.sh
```
It installs dependencies, creates your .env, sets up local Whisper, and launches the app.
3

Add your Gemini key

On first launch the Settings window opens automatically. Get a free key from Google AI Studio and paste it in.