Axeom/tribe-engage

Fix: install spaCy en_core_web_lg model required by tribev2

Fix: handle torch tensors from real model + surface all error details

Fix timeout: frontend uploads directly to HF Space, saves via Worker

Fix: install uv (tribev2 needs uvx for subprocess)

Debug: include error details in prediction failure response

Fix: tribev2 import path + add HF_TOKEN for gated model access + better error logging

Fix: install tribev2 from GitHub repo, not PyPI

Fix: Force torch+tribev2 install on GPU tier (no silent fallback to demo)

Simplify to pure brain activity viewer

Add Brain Activity Explorer: synced video + region heatmap timeline

Fix insight engine: hook/closing windows for short videos, smarter dimension thresholds

Add actionable creator insights UI

Rename to Quantis by Axeom — remove all TRIBE branding and emojis

Light-mode Palantir redesign + fix HF Space access

Fix HF build: split requirements so CPU tier succeeds (demo mode)

Redesign UI with Palantir-style aesthetic

Migrate frontend from CRA to Vite, deploy to Cloudflare Pages

Add D1 database_id to wrangler.toml

Move auth + storage to Cloudflare Worker with D1

Initial production release — TRIBE Engage

initial commit