Spaces:
Sleeping
30-in-15 · #17 — DIY Creator (Multi-Agent Team) (build-in-public post)
Attach: a screen-record of the team lane building — Researcher brief → Writer draft → Critic scores & sends back → revised → final guide. Demo: https://diy.gritai.solutions · Repo: https://github.com/GritAI-Labs/multi-agent-diy
🛠️ Project #17 of my "30 AI Projects in 15 days": a multi-agent team you can watch work.
You name something you want to build — a raised garden bed, a floating shelf, a cold frame — and three agents hand off to each other to produce a vetted build guide:
🔍 a Researcher gathers the materials, tools, and safety ✍️ a Writer drafts a step-by-step guide from that brief 🧐 a Critic scores the draft, flags what's wrong, and sends it back — the Writer revises, the Critic re-checks, and it loops until it passes.
The interesting part isn't that it's "multi-agent" — it's the two design choices under it. The critic is independent (ideally a different model), so it's not the author grading its own work — the "diamond." And the pass/revise decision is made in code from issue severity, not by asking the model "is this good?" — so it can't rubber-stamp itself. A small local model reviewing itself is honestly noisy, and the app says so rather than faking a clean pass.
Runs on a local model by default (free, private) or a hosted one. You watch the whole team work, live.
⭐ https://github.com/GritAI-Labs/multi-agent-diy
X / Twitter version (≤280)
🛠️ #17 of my 30 AI projects in 15 days: a multi-agent team you watch work.
Name a DIY project → 🔍 Researcher gathers materials/tools/safety → ✍️ Writer drafts a guide → 🧐 Critic scores it & sends it back to revise, looping til it passes. Independent critic; the gate is code, not vibes.
→ https://github.com/GritAI-Labs/multi-agent-diy
Permalinks: X: _____ · LinkedIn: _____