cue  by Blueturboguy07

AI copilot for macOS assisting live meetings and coding

Created 1 week ago

New!

710 stars

Top 47.5% on SourcePulse

GitHubView on GitHub
Project Summary

An open-source AI copilot for macOS, cue provides real-time assistance during meetings and coding tasks by processing screen, microphone, and meeting audio inputs. It aims to be a private, self-hosted alternative to commercial tools like Cluely, allowing users to leverage their own API keys from providers such as OpenAI, Anthropic, or Google Gemini.

How It Works

Built as an Electron application, cue captures screen content, microphone audio ("You"), and system audio output ("Them") as distinct data streams. These inputs are processed locally, with audio transcribed via Whisper or Gemini, and then fed to a user-selected AI model for real-time responses. The application's "invisibility" in screen shares is achieved using the macOS setContentProtection(true) window flag, which requests exclusion from screen capture streams.

Quick Start & Requirements

  • Installation: Download cue-mac.zip from the Releases page and drag cue.app to Applications. For developers, clone the repository (git clone https://github.com/Blueturboguy07/cue.git), navigate to the directory, and run npm install followed by npm start.
  • Prerequisites: macOS. For source builds, Node.js 18+ is required. Users must grant Microphone and Screen Recording permissions to cue. An API key from OpenAI, Anthropic, or Google Gemini is necessary.
  • Setup: The first launch requires a one-time setup: right-click cue.app and select "Open," or use the Terminal command xattr -cr /Applications/cue.app. A built-in tutorial guides through granting permissions and adding the AI key.
  • Zoom Specific: For Zoom, enable "Advanced capture with window filtering" in Share Screen settings.
  • Links: GitHub Repository

Highlighted Details

  • Offers multiple interaction modes: "Assist" (screen + conversation), "What should I say?" (meeting audio + mic), "Solve a coding problem" (screen only), and general Q&A.
  • Features a "Smart toggle" for selecting between more thorough or faster AI models.
  • The overlay panel is transparent and click-through, designed not to obstruct underlying applications.
  • All AI processing occurs locally, with API keys stored only on the user's machine.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), or roadmap are provided in the README.

Licensing & Compatibility

The project is licensed under GPL-3.0-or-later. This copyleft license requires derivative works to be distributed under the same terms, potentially restricting integration into closed-source or commercial products without open-sourcing modifications.

Limitations & Caveats

The "invisibility" feature is best-effort and not guaranteed, particularly on macOS 15.4+, where modern capture tools may bypass the protection. Users are solely responsible for ensuring their use complies with platform rules and consent laws. Rebuilding the application resets macOS permissions, requiring re-granting. Certain API keys (e.g., OpenAI chat-only) may lack necessary capabilities for all features.

Health Check
Last Commit

6 days ago

Responsiveness

Inactive

Pull Requests (30d)
9
Issues (30d)
7
Star History
713 stars in the last 12 days

Explore Similar Projects

Feedback? Help us improve.