Creator Guild AI OS

A Next.js + TypeScript + Supabase AI operating system for managing a live-streaming creator guild.

Setup

1. Install dependencies

npm install

2. Configure environment variables

Create a .env.local file in the project root:

# Supabase
NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_supabase_service_role_key

# AI Providers — only the key for providers you actually use is required
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
GEMINI_API_KEY=AIza...

# Optional: use a Gemini relay/proxy if Google APIs are blocked
# from your server network (for example in restricted regions).
GEMINI_BASE_URL=https://your-proxy-endpoint.com

A missing key for a provider only causes an error when that provider is selected for a task. Unused providers are completely ignored. If Gemini requests fail with network errors such as fetch failed or ENOTFOUND, set GEMINI_BASE_URL to a reachable relay endpoint.

3. Run migrations

Apply all migrations in supabase/migrations in filename order to the same Supabase project referenced by NEXT_PUBLIC_SUPABASE_URL.

4. Seed data

# via Supabase CLI
supabase db seed

Or run supabase/seed.sql in the Supabase SQL editor.

5. Run the dev server

npm run dev

---

Agent Model Configuration

Each agent can independently use a different AI provider and model.

Model resolution priority

1. Database config — agents.model_provider + agents.model_name from the database row
2. Code defaults — role-based fallback defined in src/lib/agents/model-config.ts
3. Error — if still unresolved, execution throws a clear error (no silent fallback)

There is no automatic failover across providers. If a provider is selected but its API key is missing, the task fails with a clear error message.

Supported providers

Provider	Env var	Example models
anthropic	ANTHROPIC_API_KEY	claude-sonnet-4-6, claude-opus-4-7
openai	OPENAI_API_KEY	gpt-4o, gpt-4o-mini
gemini	GEMINI_API_KEY + optional GEMINI_BASE_URL	gemini-1.5-pro, gemini-2.0-flash

Editing via the Team page

Navigate to /team. Each agent card has a Model Configuration section at the bottom with provider and model dropdowns. Changes are saved immediately to the database via PATCH /api/agents/:id.

---

Architecture

• Agents are data-driven — behavior is defined by prompt templates in the database.
• Executor (src/lib/agents/executor.ts) is a single generic function for all agents.
• Provider abstraction (src/lib/agents/providers.ts) handles Anthropic / OpenAI / Gemini.
• Model resolution (src/lib/agents/model-config.ts) applies the DB-first fallback chain.
• Structured JSON output is enforced at the prompt level and via native response_format / MIME type where supported.