Deployment¶

Deploy the seller agent using Docker locally or to AWS with CloudFormation or Terraform.

Quick Start — Docker Compose¶

The fastest way to run the full stack (app + PostgreSQL + Redis):

cd infra/docker
docker compose up

This starts:

Service	Port	Purpose
app	8000	Seller agent API
postgres	5432	Durable business data (products, deals, orders)
redis	6379	Sessions, cache, pubsub

Verify it's running:

curl http://localhost:8000/health

Environment Variables¶

The app container reads from ../../.env (project root). Key production settings:

STORAGE_TYPE=hybrid                    # Routes keys to Postgres or Redis
DATABASE_URL=postgresql+asyncpg://seller:seller@postgres:5432/ad_seller
REDIS_URL=redis://redis:6379/0
ANTHROPIC_API_KEY=sk-ant-...           # Or your chosen LLM provider key

See Configuration for the full variable reference.

Rebuilding¶

docker compose build --no-cache app
docker compose up -d

Storage Backends¶

The seller agent supports three storage modes:

Mode	`STORAGE_TYPE`	Best For
SQLite	`sqlite`	Local dev, single instance
Redis	`redis`	Fast ephemeral storage
Hybrid	`hybrid`	Production — Postgres for business data, Redis for sessions/cache

Hybrid Mode (Recommended for Production)¶

Hybrid mode routes keys by prefix:

Redis: session:*, session_index:*, cache:*, lock:*, pubsub:*, rate_limit:*
PostgreSQL: Everything else (products, deals, orders, proposals, negotiations, quotes, agents, packages)

This gives you durable storage for business data with fast in-memory access for sessions.

STORAGE_TYPE=hybrid
DATABASE_URL=postgresql+asyncpg://user:pass@host:5432/ad_seller
REDIS_URL=redis://host:6379/0

PostgreSQL Connection Pool¶

For production tuning:

POSTGRES_POOL_MIN=2    # Minimum connections (default: 2)
POSTGRES_POOL_MAX=10   # Maximum connections (default: 10)

AWS Deployment¶

Two IaC options are provided — choose based on your team's preference:

CloudFormation¶

Nested stack templates in infra/aws/cloudformation/:

cloudformation/
├── main.yaml       # Root stack (orchestrates nested stacks)
├── network.yaml    # VPC, subnets, NAT, security groups
├── storage.yaml    # Aurora Serverless v2, ElastiCache Redis
└── compute.yaml    # ECS Fargate, ALB, CloudWatch, IAM

Deploy:

# Upload nested templates to S3
aws s3 sync infra/aws/cloudformation/ s3://your-bucket/cf-templates/

# Create the stack
aws cloudformation create-stack \
  --stack-name ad-seller-prod \
  --template-url https://your-bucket.s3.amazonaws.com/cf-templates/main.yaml \
  --parameters \
    ParameterKey=Environment,ParameterValue=production \
    ParameterKey=AnthropicApiKey,ParameterValue=sk-ant-... \
    ParameterKey=ContainerImage,ParameterValue=123456789.dkr.ecr.us-east-1.amazonaws.com/ad-seller:latest \
  --capabilities CAPABILITY_NAMED_IAM

Terraform¶

Modular Terraform in infra/aws/terraform/:

terraform/
├── main.tf
├── variables.tf
├── outputs.tf
├── terraform.tfvars.example
└── modules/
    ├── network/
    ├── storage/
    └── compute/

Deploy:

cd infra/aws/terraform
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars with your values

terraform init
terraform plan
terraform apply

AWS Architecture¶

Both options deploy the same architecture:

Compute: ECS Fargate (serverless containers, auto-scaling)
Database: Aurora Serverless v2 PostgreSQL (auto-scaling 0.5–4 ACU)
Cache: ElastiCache Redis (t3.micro)
Networking: VPC with public/private subnets across 2 AZs
Load Balancer: Application Load Balancer with HTTPS
Secrets: SSM Parameter Store (SecureString)
Logging: CloudWatch Logs

Building the Container Image¶

For ECR deployment:

# Build
docker build -t ad-seller -f infra/docker/Dockerfile .

# Tag and push to ECR
aws ecr get-login-password --region us-east-1 | docker login --username AWS --password-stdin 123456789.dkr.ecr.us-east-1.amazonaws.com
docker tag ad-seller:latest 123456789.dkr.ecr.us-east-1.amazonaws.com/ad-seller:latest
docker push 123456789.dkr.ecr.us-east-1.amazonaws.com/ad-seller:latest