Project Overview
Relevant source files
The following files were used as context for generating this wiki page:
HEX-YT-INTEL is an elite YouTube content synthesis and intelligence engine designed to transform long-form video content into structured, actionable intelligence databases. The system utilizes the Ultimate Content Intelligence & Implementation System (UCIS) v5.1 to deconstruct transcripts into 11 distinct knowledge dimensions, ranging from executive summaries to commercial viability assessments.
The project is built on a Hybrid Edge Architecture, coordinating high-security authentication via Vercel with high-performance, high-latency LLM synthesis via Cloudflare Workers. It targets power users such as content creators, indie makers, and consultants, providing them with a "synthesis console" that maps video insights directly into a persistent knowledge graph.
Sources: README.md:1-5, CLAUDE.md:16-22, design-system/design-brief.md:4-10, web/lib/prompts/ucis-v5.1.ts:5-15
System Architecture
The platform implements the Hybrid Edge Symphony, a multi-cloud architecture designed to overcome standard serverless execution limits and ensure data integrity.
Hybrid Edge Flow
The architecture distributes responsibilities across three main layers:
- Vercel Bouncer: Handles high-security authentication (Supabase) and atomic quota enforcement (Upstash Redis) with a ~8s execution window.
- Cloudflare Streamer: Manages direct browser-to-edge SSE connections for high-latency LLM synthesis tasks, allowing for up to ~58s of processing time.
- HMAC Persistence: Utilizes cryptographically signed server-to-server (S2S) database writes to guarantee data integrity between the edge worker and the Supabase backend.
Sources: README.md:12-18, CLAUDE.md:16-22, GEMINI.md:46-52
Architecture Diagram
The following diagram illustrates the request flow and component interaction within the Hybrid Edge Symphony.
The diagram shows the request lifecycle from initial authentication at Vercel to streaming synthesis via Cloudflare and final persistence in Supabase.
Sources: README.md:12-23, CLAUDE.md:16-22, GEMINI.md:46-52
Core Intelligence: UCIS v5.1
The heart of the project is the Ultimate Content Intelligence & Implementation System (UCIS) v5.1. This framework deconstructs content based on "Transcript Absolutism"—operating in a strictly deterministic, closed-universe sandbox where only the provided transcript is used as a source of truth.
The 11 Knowledge Dimensions
The system generates a multi-dimensional intelligence product containing:
- Dimension 1: Apex Intelligence: 60-second ROI summary.
- Dimension 8: Semantic & Knowledge Graph: Structured nodes for RAG ingestion.
- Dimension 11: Commercial Yield: Monetization profiling across AdSense, sponsorships, and leads.
Sources: web/lib/prompts/ucis-v5.1.ts:18-35, supabase/migrations/20260611142500_add_prompt_config.sql:10-50
Persona-Weighted Intelligence
Analysis is weighted toward five specific target personas to ensure practical utility.
| Persona ID | Label | Weight | Focus Areas |
|---|---|---|---|
| P1 | Content Creator | 50% | RPM optimization, growth velocity, AdSense |
| P2 | Indie Maker | 25% | Lead generation, service positioning, scalability |
| P3 | Consultant | 15% | Advisory retainers, fractional arrangements |
| P4 | Researcher | 5% | Academic positioning, grant/foundation support |
| P5 | Product Manager | 5% | Internal alignment, thought leadership |
Sources: web/lib/prompts/ucis-v5.1.ts:70-80, web/lib/prompts/ucis-v5.1.ts:265-290
Design System
The project uses a premium, dark-themed design system described as a "technical instrument register." It emphasizes high data density and precision while reducing cognitive load through structured layouts.
Visual Foundations
- Color Palette: Primarily Obsidian (
#11141D) and Slate backgrounds, with Electric Cyan (#06B6D4) as the exclusive accent color for active or streaming states. - Typography:
Interfor body text andJetBrains Monofor metadata and technical strings. - Layout: Uses an "asymmetric bento" grid with span weights and signature scope-bracket framing.
Sources: design-system/design-brief.md:21-35, CLAUDE.md:81-90
Data Management & Infrastructure
Technical Stack
The project enforces a "Frozen Stack Protocol" to maintain high performance and bundle efficiency.
| Category | Technology | Version / Note |
|---|---|---|
| Framework | Next.js | 16.2.6 (React 19) |
| Language | TypeScript | 5.6.2 |
| Styling | Tailwind CSS | v4 theme layer |
| Database | Supabase | Auth and Persistence |
| Caching | Upstash Redis | Rate limiting & KV cache |
| Package Mgr | pnpm | 11.1.3 |
Sources: GEMINI.md:15-30, CLAUDE.md:92-105
Core Operational Laws
The system operates under strict engineering constraints to ensure reliability:
- Cache Hit Circuit: All requests first check Supabase for existing
video_id/user_idmatches to return cached markdown instantly at $0 cost. - Stratified Dual-Timeouts: Implements a 3s connection timeout and a 25s (Vercel) / 90s (Worker) token streaming window.
- Root Restriction: The repository root is limited to a maximum of 4 Markdown files (
CLAUDE.md,GEMINI.md,README.md,AGENTS.md).
Sources: CLAUDE.md:5-14, GEMINI.md:31-35, GEMINI.md:71-80
Summary
HEX-YT-INTEL is a sophisticated AI synthesis platform that prioritizes technical accuracy and architectural resilience. By combining the UCIS v5.1 framework with a Hybrid Edge Architecture, the system provides high-fidelity content deconstruction while overcoming the performance limitations typical of serverless AI applications.