Every dev conversation contains decisions. Requirements shift, constraints surface, tradeoffs get made. But most of it disappears into Slack threads and meeting notes that nobody reads again.
I built Forge to fix that.
Forge is persistent project intelligence. It takes conversations — standups, design reviews, planning sessions — and converts them into structured decisions, constraints, and artifacts that your project can actually use.
The flow
conversation → structured meaning → decision → artifact → system change
A builder talks. The product thinks. The system evolves.
What it does
- Ingests conversation transcripts (meetings, chats, design sessions)
- Extracts structured entities: decisions, constraints, requirements, risks
- Tracks decision history with full context and rationale
- Generates artifacts: specs, ADRs, requirement docs
- Detects when new conversations contradict previous decisions
- Maintains a living project knowledge base
How it works
You feed Forge your conversations (paste, API, or file watch). It uses LLMs to extract meaning, then structures that meaning into a queryable knowledge base. When you need to know "why did we decide X?" or "what constraints apply to Y?" — Forge has the answer with source citations.
Quick start
npx @gzoo/forge init
npx @gzoo/forge ingest ./meeting-notes/
npx @gzoo/forge query "what did we decide about the auth flow?"
Why this matters
Teams make hundreds of decisions a week. Most are never recorded. When someone asks "why is it like this?" six months later, nobody remembers. Forge captures the reasoning automatically so your project has a memory.
Open source, MIT licensed.
GitHub: github.com/gzoonet/forge
npm: npmjs.com/package/@gzoo/forge
If you've ever wasted time re-litigating a decision that was already made — this is for you. Feedback welcome.
