SparkGen Deploy

Deploy the agent server in the specified mode.

Dynamic Context

Before deploying, gather current state:

1. •Read .env (if exists) to check current DEPLOYMENT_MODE and LLM_PROVIDER
2. •Check if server is already running: curl -sf http://localhost:8000/health
3. •For docker mode: check Docker is running with docker info
4. •For local mode: check Ollama with curl -sf http://localhost:11434/api/tags

Deployment Modes

Local (/sparkgen-deploy local)

1. •Verify Ollama is running: curl -sf http://localhost:11434/api/tags
   • •If not: instruct user to run ollama serve and ollama pull llama3.2:3b
2. •Source environment: set -a && source .env.local && set +a
3. •Start server: uvicorn app.api:app --host 0.0.0.0 --port 8000 --reload
4. •Verify: curl -sf http://localhost:8000/health
5. •Print summary: endpoint URL, provider, model, available agents

Docker (/sparkgen-deploy docker)

1. •Verify Docker is running: docker info
2. •Copy env file: cp .env.docker .env
3. •Start stack: docker compose up --build -d
4. •Wait for health: poll curl -sf http://localhost:8080/health (up to 60s)
5. •Check all services: docker compose ps
6. •Print summary: agent URL (:8080), DynamoDB Local (:8000), Redis (:6379)

AWS (/sparkgen-deploy aws [--target eks|lambda])

1. •Verify AWS credentials: aws sts get-caller-identity
2. •Check .env.aws exists and has required vars (AWS_REGION, AWS_MODEL_ID)
3. •For EKS: run bash scripts/deploy.sh eks
4. •For Lambda: run bash scripts/deploy.sh lambda
5. •Verify deployment health via API Gateway URL from .env.aws

Error Handling

• •If port 8000 is already in use: suggest lsof -i :8000 to find the process
• •If Ollama model not found: run ollama pull <model>
• •If Docker build fails: show docker compose logs output