About Stream Manager

How it's built

Stream Manager is built with substantial help from Claude (Anthropic's AI assistant) and GitHub Copilot. I'm a front-end developer by trade with only a cursory understanding of backend systems, and building a full Electron desktop app with multi-track video processing, ffmpeg integration, OAuth flows, and cloud-aware file handling is well outside my normal scope.

Claude handled much of the architecture, implementation, debugging, cleanup, and documentation across months of iterative back-and-forth, while the product direction, UX decisions, feature scope, testing, and final calls on what ships are mine. If you'd like to see a fully hand-coded project of mine, take a look at ClpChk. If you're curious about my take on AI-assisted development, I'm happy to chat in GitHub issues or on Discord.

Portable and self-contained

Stream Manager is a single portable executable. There is no installer and nothing to extract: download the .exe and run it from anywhere, including an external drive. You can move it freely without breaking anything.

Your settings are stored in your Windows AppData folder, and each stream's data is kept right next to its files. Your library stays portable and self-describing even if you relocate the app or move your recordings to another drive or machine.

Your data

Stream Manager runs locally and needs no account to use. Your recordings, metadata, settings, and any API keys stay on your machine, and the keys and account tokens are stored encrypted. Connecting YouTube, Twitch, or Claude is optional, and data is only sent to those services when you explicitly ask for it, such as publishing your metadata, running the relay, or requesting an AI suggestion. Even then, only the specific data that action needs is sent.

How the app behaves around your data is not just a description of how it works today. The design principles set out the rules it is built to, written so you can check each one against the app or its source.

A note on Windows prompts and code signing

Stream Manager is not signed with a commercial code-signing certificate. Those certificates are expensive and mostly serve large vendors, so a free, open source project ships unsigned for now. The first time you run it, Windows SmartScreen may show a "Windows protected your PC" warning. Click "More info", then "Run anyway" to launch it.

Because the whole project is open source, you can review exactly what it does on GitHub before running it. You can even have an AI assistant like Claude or ChatGPT review the code for you. If you want to build it yourself, see the GitHub repository for instructions.

Releases are now built by an automated pipeline on GitHub's own infrastructure, straight from the tagged source, so no personal machine touches the file you download. The build only publishes if the code passes its typecheck and lint, every release's full build log is public under the repository's Actions tab, and GitHub lists a SHA-256 checksum beside the download. None of this removes the SmartScreen warning, but it does mean you can trace the file you downloaded back to the exact code that produced it.

If you turn on Stream Relay, you may see a second Windows prompt. The relay runs a small local RTMP server, using the ffmpeg build bundled with the app, that listens for your streaming software. The first time a new version starts with the relay enabled, Windows Defender Firewall asks whether to allow ffmpeg.exe to accept incoming connections, and allowing it on private networks is the expected choice. This is a firewall permission, not an elevation: ffmpeg never runs with administrator rights, and the admin consent on that dialog only covers creating the firewall rule. The prompt returns once per new version because each version's files live at a new path, and that ffmpeg binary ships inside the same CI-built, checksummed release as the app itself. If you never enable Stream Relay, you will not see it at all.