v0.1.0 — an AI team whose work you can audit (feedback wanted)

[FORIFOR]

Agent Team v0.1.0 is out. One request → a Master plans → Researcher / Builder / Reviewer actually do the work → you get artifacts plus the real bot-to-bot messages, a timeline, and checks bound to each artifact revision.

• Site + 37s walkthrough: https://forifor.github.io/Multibot/
• Release: https://github.com/FORIFOR/Multibot/releases/tag/v0.1.0

What I'd like feedback on:

1. The review → revise → re-review loop bound to artifact revisions — does the "old check never applies to new content" rule hold up for your tasks?
2. "Fewer messages, more evidence": workers get their task, input artifact refs and their own inbox, not the whole history. Where does that break?
3. Per-bot endpoint/model/prompt with no silent fallbacks — which providers should the probe cover next?

Honesty note: the demo drives the real UI/runtime with the scripted test provider (labelled on screen). Real-model runs are not yet published; backend/scripts/smoke_real_llm.py prints the evidence if you run it with a key.