SKATE

I run a lot of workshops for work and I got tired of ending up with a giant transcript and no good way to answer the only question that matters later. “Why did we decide to do this?” Granola, Otter, etc are fine for meeting notes. I just wanted something more specific to workshops and more customizable. So I built SKATE using Claude Code. It lets me mark observations, pains, questions, actions, decisions, quotes, all that stuff while the workshop is happening, then connect them with actual relationships like this caused that, this supports that, this contradicts that. The goal is not "summarize my meeting." The goal is more like "show me how we got from what somebody said in the room to the thing we eventually recommended." That sounds boring until six weeks later when somebody asks why the hell you made a decision and everyone starts digging through notes like raccoons in a dumpster. The notes are just Markdown files much like Obsidian. No database. No magic memory box. If the app dies you still have the files. I also wanted local transcription because I was getting tired of meeting bots. So it can record PC audio and mic, transcribe locally, and it does not need to send a little AI intern into the call to sit there silently collecting your sins. Then I decided the software needed hardware. So now there is a 3D printed microphone puck. STL in the repo. And a custom Stream Deck. All config on there too. This was supposed to be a small project…whoops. It is no longer a small project. I built most of this with Claude Code. A lot of it was basically me having an idea, explaining it badly to Claude, building a version, breaking it, figuring out why it was stupid, and doing it again. Everyone has mind map software right now and I wanted to Skate the mindmap. Props to Claude Design for making the video above. Claude Code helped me work through the app itself, the local transcription, Stream Deck stuff (Even did the button logos), the MCP server, tests, and a truly unreasonable amount of debugging. It was especially useful once the project got big enough that I could point it at the repo and say "ok what did I break this time" instead of trying to stuff half the codebase into a chat. Sometimes Claude and I were both extremely confident and completely wrong but we got through it :) I assume this is what pair programming is supposed to feel like. There is also an MCP server now so Claude and ChatGPT can query the workshop memory without me dumping the whole thing into context every time. Retrieval stays bounded as the vault gets bigger, which was one of the things I cared about when building it. The whole thing is free and open source under MIT. Clone it, use it, break it, whatever. Please give it a like would mean the world to me. A few things are still rough. Windows only right now. Single user. Probably some other dumb stuff I have stared at for so long I can no longer see it which is why I’m posting here. If you actually run workshops, discovery sessions, process improvement stuff, consulting, whatever, I would love feedback. What is useful? What is stupid? What is missing? What would make you never touch this again? Repo: https://github.com/SixSigmaEngineer/skate-workshop-os?utm_source=chatgpt.com Did I make a video of Denzel talking about this slop? Yes. https://youtu.be/MjWGChZChCc?si=0nRVIBKTvmBASM0J I have made peace with the person I have become.