Configuration

All settings can be configured via CLI flags, environment variables, or a .env file.

Control Plane

# Server binding
fabric serve --host 0.0.0.0 --port 8080 --grpc-port 9090

# Database
fabric serve --database-url postgres://user:pass@host/db

# Authentication
fabric serve --jwt-secret $SUPABASE_JWT_SECRET --fabric-env production

# AI Providers
fabric serve --openai-api-key sk-... --anthropic-api-key sk-ant-...
fabric serve --ollama-enabled --ollama-url http://gpu-server:11434
fabric serve --comfyui-enabled --comfyui-url http://localhost:8188
fabric serve --onnx-model-dir /models --candle-enabled

Execution Plane

# Worker configuration
fabric executor --concurrency 8 --worker-id gpu-worker-1
fabric executor --poll-interval 500 --lease-seconds 60 --max-retries 5

Env Var	Default	Description
`WORKER_ID`	random UUID	Unique worker identity
`WORKER_CONCURRENCY`	4	Max steps to claim per poll cycle
`WORKER_POLL_INTERVAL`	500ms	How often to check for work
`WORKER_LEASE_SECONDS`	30	Lease TTL before auto-expiry
`WORKER_MAX_RETRIES`	3	Max retry attempts per step

Environment Variables

Server

Variable	Default	Description
`HOST`	`127.0.0.1`	HTTP bind address
`PORT`	`3001`	HTTP port
`GRPC_PORT`	`3002`	gRPC port
`DATABASE_URL`	—	Postgres connection string
`FABRIC_ENV`	`development`	Environment (`development` or `production`)
`RUST_LOG`	`info`	Log level filter

Authentication

Variable	Default	Description
`JWT_SECRET`	—	Supabase JWT HS256 secret
`FABRIC_ENV`	`development`	Set to `production` to disable dev headers

AI Providers

Variable	Default	Description
`OPENAI_API_KEY`	—	Enables OpenAI provider
`OPENAI_BASE_URL`	`https://api.openai.com/v1`	Custom OpenAI-compatible endpoint
`ANTHROPIC_API_KEY`	—	Enables Anthropic provider
`OLLAMA_ENABLED`	`false`	Enable Ollama provider
`OLLAMA_URL`	`http://localhost:11434`	Ollama server URL
`WHISPER_URL`	—	Whisper server URL
`COMFYUI_ENABLED`	`false`	Enable ComfyUI provider
`COMFYUI_URL`	`http://localhost:8188`	ComfyUI server URL
`ONNX_MODEL_DIR`	—	ONNX model directory
`CANDLE_ENABLED`	`false`	Enable Candle provider

Infrastructure (Docker Compose)

just infra-up          # Full stack (Postgres + Ollama + Whisper + ComfyUI)
just infra-up-db       # Postgres only
just infra-down        # Stop everything
just infra-reset       # Fresh start (delete volumes)
just infra-psql        # psql shell
just infra-migrate     # Run migrations
just ollama-pull <model>  # Pull an Ollama model

Docker Compose profiles:

default — Postgres
ai — Postgres + Ollama + Whisper + ComfyUI
full — Everything