klisk serve

Start the production server with chat UI and API endpoints

Usage

klisk serve [name_or_path] [options]

Description

Launches the production server with a public-facing chat UI and REST API. Optimized for deployment environments.

Arguments

name_or_path · string · default: "."

Project name or path to serve.

Options

--port · integer

Port to bind the server to. Use -p as shorthand.

Default: Reads from $PORT environment variable, falls back to 8080

--host · string · default: "0.0.0.0"

Host address to bind to. Use -h as shorthand.

Default: 0.0.0.0 (all interfaces)

Server Endpoints

The production server exposes:

Chat UI

http://localhost:8080

Web interface for interacting with your agent.

Chat API

http://localhost:8080/api/chat

REST endpoint for programmatic access.

Health Check

http://localhost:8080/health

Returns server status for monitoring.

Examples

Default (Current Directory)

klisk serve

Output:

  Chat UI: http://0.0.0.0:8080
  API:     http://0.0.0.0:8080/api/chat
  Health:  http://0.0.0.0:8080/health
  Project: /path/to/project

Specific Project

klisk serve my-agent

Custom Port

klisk serve --port 3000

Custom Host and Port

klisk serve --host 127.0.0.1 --port 5000

Using Environment Variable

PORT=9000 klisk serve my-agent

Cloud Deployment

Most cloud platforms set the PORT environment variable automatically:

# Railway, Render, etc.
klisk serve

The server will use the platform’s assigned port.

Configuration

The server loads configuration from:

1. Project config: klisk.config.yaml
2. Environment variables: .env file in the project root
3. Runtime options: --port and --host flags

Port Priority

1. --port flag (highest priority)
2. $PORT environment variable
3. Default: 8080

Environment Variables

Ensure your .env file contains necessary API keys:

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...

The server loads these on startup.

Production Considerations

Deployment Checklist

• Set real API keys in .env
• Configure appropriate host (0.0.0.0 for containers, 127.0.0.1 for local)
• Use environment variable for port (cloud platforms)
• Set up health check monitoring at /health
• Consider rate limiting and authentication

Performance

The production server:

• Uses production ASGI settings
• Disables hot reload (unlike klisk dev)
• Optimized for concurrent requests
• No debug mode or verbose logging

Security

For production deployments:

• Use HTTPS (configure at reverse proxy level)
• Implement authentication/authorization
• Set appropriate CORS policies
• Protect API keys (never commit to version control)

Differences from klisk dev

Feature	klisk dev	klisk serve
Hot reload	Yes	No
Studio features	Full development UI	Production chat UI
Port default	8000 (from config)	8080 (from $PORT or default)
Workspace mode	Supported	Single project only
Logging	Verbose	Production-level
Use case	Development	Production/staging

API Usage

Interact with the chat API programmatically:

curl -X POST http://localhost:8080/api/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Hello, agent!",
    "agent": "my-agent"
  }'

Monitoring

Use the health endpoint for uptime monitoring:

curl http://localhost:8080/health

Integrate with monitoring services:

• Uptime Robot
• Pingdom
• Cloud provider health checks (AWS, GCP, Azure)

Deployment Examples

Docker

FROM python:3.12-slim
WORKDIR /app
COPY . .
RUN pip install klisk
EXPOSE 8080
CMD ["klisk", "serve", ".", "--port", "8080"]

Railway

# Railway automatically sets $PORT
klisk serve

Google Cloud Run

See klisk deploy for automated Cloud Run deployment.

Related Commands

• klisk dev - Development server with hot reload
• klisk deploy - Deploy to Google Cloud Run
• klisk run - Run agent from CLI
• klisk check - Validate before serving