Get real-time AI captioning or note-taking support during a solo study or practice session.
Capture on-screen code and receive a solution with time and space complexity explained.
Draft a reply or summary of an ongoing conversation using recent transcript context.
Ask the overlay's composer a custom question using whatever is currently visible on screen.
| ganeshp000/prochame | blacksnowskill/comfyui-bss_flsampler | gaearon/redux-devtools-themes | |
|---|---|---|---|
| Stars | 34 | 34 | 34 |
| Language | JavaScript | JavaScript | JavaScript |
| Last pushed | — | — | 2019-11-06 |
| Maintenance | — | — | Dormant |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 2/5 | 1/5 |
| Audience | general | vibe coder | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires granting microphone and screen recording permissions, an AI provider API key, and Node.js 18 or later to build from source.
Prochame is a desktop application that displays a small, dark, semi-transparent overlay window floating on top of other windows. It captures what is on the screen and transcribes microphone or system audio, then sends that information to an AI language model and streams back suggestions in real time. The overlay is built using screen capture protection features so that, in many cases, it does not appear when the screen is shared or recorded through common tools like Zoom, Google Meet, or Microsoft Teams. The README is unusually direct about the ethics of this design. It states plainly that the invisibility feature is best effort rather than guaranteed, since phone cameras, some proctoring software, HDMI capture devices, and full display recording can still show the overlay. It lists the intended uses as personal study, accessibility support such as live captioning, solo productivity work, and learning about AI assisted workflows. It then explicitly states the tool must not be used to cheat on proctored exams or certifications, gain an unfair advantage in job interviews, break a platform's terms of service, or record and transcribe other people without their knowledge or consent. The README places responsibility for how the tool is used entirely on the person running it. Feature wise, keyboard shortcuts trigger different actions: one captures the screen and recent conversation to ask the AI for general help, another captures visible code and returns a solution with its time and space complexity, and others draft a reply to a conversation, summarize a discussion into action items, or generate follow up questions to keep a conversation going. A composer window also lets a user type free-form questions for the AI to answer using the captured context. The project is built with Electron, combined with a React and Vite based interface, and is available as a prebuilt download for Apple Silicon Macs or as an unsigned Windows installer, which triggers a Windows SmartScreen warning on first run. Building it from source requires Node.js version 18 or later and involves running a development server alongside the Electron shell. On Windows, full invisibility from screen capture depends on having a version from 2020 or later, older Windows versions instead show a black rectangle in place of the overlay. The README also covers granting microphone and screen recording permissions on both operating systems and configuring API keys, starting with OpenAI, through a settings panel in the app.
A desktop overlay app that watches your screen and microphone and streams live AI suggestions, built with screen-capture protection so it often stays hidden from calls and recordings.
Mainly JavaScript. The stack also includes JavaScript, Electron, React.
Not sufficiently specified in the provided material.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
double-check against the repo, no cap.