Free, open-source tools that close the gap between "I see the bug" and "the AI fixes it." š BugCapture ā screen + audio recording ā AI-readymd with screenshots + Whisper transcript. Drop into Claude or Copilot; the AI understands the bug immediately. š§ LearnBoard ā persistent AI memory via LEARNING.md. Sessions get smarter over time. šØ PromptBoard ā drag-and-drop canvas for building AI prompts visually. Try in-browser at machina.chat. MIT licensed, runs locally.
No reviews yetBe the first to leave a review for Machina
Maker
š
Hey PH! š I'm Alex, the developer behind Machina.
I built these tools out of frustration. I use AI agents (Claude, Copilot) every day for debugging, but I kept running into the same friction:
⢠The AI doesn't remember what it learned last session
⢠Explaining a bug in a text box loses all the visual context
⢠Assembling the right context before each session takes 10ā15 minutes
Each tool solves one specific piece of that problem. They're designed to work together, but each one is useful standalone.
The thing I'm most proud of is LearnBoard ā it turns AI sessions into a feedback loop. The agent writes lessons back to LEARNING.md, and starts the next session already knowing your stack, preferences, and past mistakes. After 4 months of daily use, my AI makes suggestions my junior self would've spent hours googling.
Happy to answer anything about the architecture, the voice cascade implementation, or why I chose local Whisper over cloud APIs. Ask me anything! š