Context stacking for voice transcription, screenshots, and portable commands — made for builders and engineers and ready for Cursor, Claude Code, or any input field. The current meta for talking to models is "tell it more." So we copy and paste a lot. Screenshots, logs, docs, voice transcripts. It's tedious. Field Theory makes context management fast. Hotkeys for everything. Local transcription. Voice commands. Image and text stacking. And portable commands you can invoke anywhere.
Hey everyone, Andrew here.
I built Field Theory because I got tired of copy-pasting context into AI tools all day -- the models are better when you give them more but the process of giving them more can be tedious.
With context windows getting larger, I think we need better tools for moving information around.
Field Theory started to solve a personal annoyance I had with my Bose headphones stealing the microphone from a dedicated mic. It snowballed into a replacement for all my context engineering tools (whispr, alfred, screenshotting, etc). Each feature is a different way to stack context quickly.
A few things I'm proud of:
- Portable commands
- Voice commands
- The Librarian
- Native support for CLIs or a multi-modal IDE
- Crazy window management on Mac
This is an indie project. I'm a solo dev and I use this thing 10 hours a day. If you have feedback or feature requests, let me know and thanks for trying it out.
Loot Character