OpenCluely

A free, open-source AI copilot that sees your screen, hears your meetings, and stays invisible while it does it.

Bring your own AI key — OpenAI · Anthropic · Gemini · Mistral · NVIDIA · Ollama · OpenRouter · Custom

---

What it does

A small glass panel floats over everything on your screen. It reads three inputs — your screen, your microphone, and the audio from your call — and uses any AI model you connect to help you in real time. It's excluded from screen-share and screen-recording capture on macOS, so it's visible to you and invisible to everyone else on the call. On Windows, position it in a corner where it won't obstruct your work.

Feature	Windows	macOS	What it uses
Assist — do the smart thing right now	Ctrl + Enter	⌘ + ↵	screen + conversation
What should I say?	button	button	meeting audio
Follow-up questions	button	button	conversation
Recap — catch up a late joiner	button	button	conversation
Solve coding problem	Ctrl + H	⌘ + H	screen only
Settings	Ctrl + ,	⌘ + ,	—
Quit	Ctrl + Shift + X	⌘ + ⇧ + X	—

Quick start

Windows

1. Go to Releases and download OpenCluely-Windows-x64.zip
2. Extract the zip and run OpenCluely.exe
3. Open Settings (Ctrl + ,), pick a provider, paste your API key
4. Press Ctrl + Enter to assist with whatever's on screen

macOS

1. Download from Releases or run from source:

   git clone https://github.com/rahulcvwebsitehosting/OpenCluely.git
   cd OpenCluely
   npm install && npm start

2. Open Settings (⌘ + ,), pick a provider, paste your API key
3. Press ⌘ + ↵ to assist with whatever's on screen

macOS first open: Right-click OpenCluely.app → Open → click Open. If told it's damaged, run xattr -cr /Applications/OpenCluely.app in Terminal once.

Building from source (any platform)

git clone https://github.com/rahulcvwebsitehosting/OpenCluely.git
cd OpenCluely
npm install
npm start

Why OpenCluely?

• Free. No subscriptions, no accounts, no telemetry.
• Your keys. Bring your own API key from any provider — you pay the provider directly, nothing marked up.
• Local models. Point it at Ollama and run the whole thing offline.
• Private. Keys and settings live in a local JSON file. No servers, no analytics, no data leaves your machine except to the AI provider you chose.
• Custom. Add any OpenAI-compatible endpoint.
• Open source. GPL-3.0. Read the code, fork it, change it.

Getting API keys

Provider	How to get a key
OpenAI	Go to platform.openai.com/api-keys, click Create new secret key. For listening features the key must have Whisper/audio access.
Anthropic	Go to console.anthropic.com, sign up, navigate to API Keys and create one.
Gemini	Go to aistudio.google.com/apikey, click Get API key. One key works for both chat and transcription.
Mistral	Go to console.mistral.ai, create an account, go to API Keys and generate a key.
NVIDIA	Go to build.nvidia.com, sign in, select any model and click Get API Key.
Ollama	No key needed. Download Ollama, run it locally, pull a model (e.g. ollama pull llama3.2).
OpenRouter	Go to openrouter.ai/keys, sign in and create a key. Single key gives access to 400+ models.
Custom	Any OpenAI-compatible endpoint. Enter the base URL, model name, and API key in the Custom provider settings.

How it stays invisible

• macOS: Uses setContentProtection to exclude its window from screen capture — the OS-level API that keeps DRM content hidden also hides OpenCluely from Zoom, Meet, OBS, and screen-recording software. You see it; a screen share doesn't.
• Windows: Uses standard window management. OpenCluely stays on top of other windows but may appear in screen recordings. For the best experience in meetings, position the panel in a corner or use Zoom's "Advanced capture with window filtering" setting (Zoom → Settings → Share Screen → Advanced → Screen capture mode).

A note on how you use this

OpenCluely is a general-purpose "see my screen, hear my call" AI tool — it doesn't know or care whether you're using it to keep pace in a fast-moving standup, practice for an interview, or get a live assist during a call. That judgment call is yours. Using it to misrepresent your own skills or knowledge in a context where you're expected to demonstrate them yourself (a graded exam, a job interview you're expected to pass unaided) is on you, not the tool.

Contributing

Issues and PRs welcome. The codebase is small and readable on purpose:

main.js         → Electron main process, IPC, window/overlay behavior
preload.js      → renderer bridge
src/llm.js      → provider abstraction (OpenAI/Anthropic/Gemini/Mistral/NVIDIA/Ollama/OpenRouter)
src/prompts.js  → the actual prompts behind each mode
src/stt.js      → speech-to-text
src/screen.js   → screenshot capture
src/store.js    → local settings persistence
renderer/       → UI

Creator

Built by Rahul Shyam

Platform	Link
LinkedIn	https://linkedin.com/in/rahulshyamcivil
X / Twitter	https://x.com/RahulShyamCV
Threads	https://threads.com/@rahulcvjps
GitHub	https://github.com/rahulcvwebsitehosting

License

GPL-3.0-or-later — free to use, modify, and share.