kilo-proxy 1.1.0

Fully OpenAI-compatible API proxy for Kilo

Kilo Proxy

A fully OpenAI-compatible API proxy for Kilo.

Features

• Full OpenAI API compatibility - chat completions, completions, embeddings, models
• All OpenAI parameters supported - max_tokens, temperature, tools, stream, response_format, etc.
• Streaming SSE support - real-time streaming responses
• Background server mode - run as a daemon process
• Token authentication management - easy auth configuration
• Cross-platform - Windows, macOS, Linux

Installation

pip install -e .

Or install from source:

pip install .

CLI Commands

Authentication

Set your Kilo authentication token:

kilo-proxy auth [token]

If no token is provided, anonymous access is used by default.

Start Server (Background)

Start the server in background mode:

kilo-proxy start [--host HOST] [--port PORT]

Options:

• --host - Host to bind to (default: 127.0.0.1)
• --port - Port to listen on (default: 5380)

Stop Server

Stop the background server:

kilo-proxy stop

Run Server (Foreground)

Run the server in foreground mode:

kilo-proxy proxy [--host HOST] [--port PORT]

Options:

• --host - Host to bind to (default: 127.0.0.1)
• --port - Port to listen on (default: 5380)

Server Status

Check the current server status:

kilo-proxy status

List Models

List all available models:

kilo-proxy models

Broke Mode (Toggle Free Models)

Toggle broke mode (only show free models):

kilo-proxy broke          # Toggle broke mode
kilo-proxy broke on       # Enable broke mode
kilo-proxy broke off      # Disable broke mode
kilo-proxy broke --list   # List all free models

Configuration Wizard

Interactive configuration:

kilo-proxy config         # Run configuration wizard
kilo-proxy config --show  # Show current config

Install OpenCode Integration

Install and configure OpenCode to use Kilo models:

kilo-proxy install opencode

This interactive wizard will:

1. Find or create your OpenCode config
2. Ask if you're broke (free models only)
3. Let you select models for all or per agent/category
4. Optionally install Oh My OpenCode support
5. Map Oh My OpenCode models to Kilo models
6. Write the configuration files

Show Configuration

Display the current configuration:

kilo-proxy config-show

Autolaunch (Run on Boot)

Install the proxy to start automatically on system boot:

kilo-proxy autolaunch

Remove the proxy from boot:

kilo-proxy unautolaunch

Works on Windows (Task Scheduler), macOS (LaunchAgents), and Linux (systemd user services).

API Endpoints

Method	Endpoint	Description
GET	/v1/models	List all available models
GET	/v1/models/{model_id}	Get a specific model
POST	/v1/chat/completions	Create a chat completion
POST	/v1/completions	Create a text completion
POST	/v1/embeddings	Create embeddings
GET	/v1/engines	List engines (legacy)
GET	/health	Health check endpoint

Usage Examples

List Models

curl http://localhost:5380/v1/models

Chat Completion (Non-streaming)

curl http://localhost:5380/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "z-ai/glm-5:free",
    "messages": [
      {"role": "user", "content": "Hello, how are you?"}
    ],
    "max_tokens": 100,
    "temperature": 0.7
  }'

Chat Completion (Streaming)

curl http://localhost:5380/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "z-ai/glm-5:free",
    "messages": [
      {"role": "user", "content": "Tell me a short story"}
    ],
    "stream": true,
    "max_tokens": 500
  }'

Configuration

File	Location
Config file	~/.kilo-proxy/config.json
PID file	~/.kilo-proxy/server.pid

Supported Models

Popular free models available through Kilo:

• z-ai/glm-5:free
• minimax/minimax-m2.5:free
• deepseek/deepseek-chat:free
• meta-llama/llama-3-8b-instruct:free
• qwen/qwen-2-7b-instruct:free

Run kilo-proxy models to see all available models.

Requirements

• Python 3.9+

License

MIT

Author

• lousybook94 (notlousybook)
• GitHub: https://github.com/notlousybook
• Email: [e-mail]
• Location: Pluto, the planet