Hey IH π
I'm a developer who uses AI agents (Claude, Copilot) every day for debugging. Over time I kept running into the same three pain points:
1. The AI forgets what it learned last session β I had to re-explain context every single time
2. Describing a bug in a text box loses all visual context β the AI can't see what I'm seeing
3. Assembling a good prompt is manual and repetitive β copy-paste from docs, stack traces, notes
So I built Machina: a suite of open-source local tools designed to fix exactly these problems.
**What's in Machina right now:**
- **BugCapture** β records your screen + audio and exports a `.md` file with a screenshot and Whisper transcript, ready to paste into any AI chat
- **ContextForge** β generates a full pre-session briefing from your git diff, SSH log, and BugCapture output
- **LearnBoard** β a UI for `LEARNING.md`, a persistent file your AI agent reads at session start so it never forgets your project conventions
- **PromptBoard** β a drag-and-drop canvas for building structured AI prompts from reusable blocks
- **Transcriber** β local Whisper service for voice input in Firefox/Brave where Web Speech API is blocked
Everything runs locally, no cloud, no subscriptions. The web versions (BugCapture and LearnBoard) work directly from machina.chat without installing anything.
**Current traction:**
- Just launched a few weeks ago
- ~5 upvotes on Product Hunt
- First blog posts getting some organic search traffic
- GitHub repo open at github.com/machina-tools/machina
Happy to answer any questions or hear feedback from builders who use AI for development. What's your biggest friction point when working with AI coding assistants?