turtles connect llm

Configure LLM connections for Turtle DNA compilation.

OpenTurtles uses LLMs to compile your one-sentence descriptions into executable code (Turtle DNA). You can connect multiple LLM servers and switch between them.

turtles connect llm add

Add or update an LLM server (interactive wizard)

Usage: turtles connect llm add <NAME>

Arguments:
  <NAME>  Server name (e.g. "openai", "deepseek", "local")

Options:
  -h, --help  Print help

Example:

$ turtles connect llm add openai

  🔌 Add LLM Server: openai

  Select provider:
    1) OpenAI
    2) Anthropic
    3) Ollama (local)
    4) Custom (OpenAI-compatible)

  Choice [1-4]: 1

  Endpoint: https://api.openai.com/v1
  API Key: sk-••••••••••••
  Model [gpt-5.4]: gpt-5.4

  ✓ LLM server 'openai' added!
    Provider: openai
    Model:    gpt-5.4
    Default:  yes ✓

  ℹ Test with: turtles connect llm test openai

Supported Providers:

All providers use the OpenAI-compatible /v1/chat/completions API format.

turtles connect llm list

$ turtles connect llm list

  LLM Servers  (2 configured)

  ● openai ★ default
    Provider: openai  Model: gpt-5.4  Endpoint: https://api.openai.com/v1
  ● deepseek
    Provider: custom  Model: deepseek-chat  Endpoint: https://api.deepseek.com/v1

turtles connect llm test

$ turtles connect llm test openai
  ✓ LLM server 'openai' connected!
    Response: Hello! How can I help you today?

turtles connect llm default

Set which LLM server is used for Turtle creation and updates.

$ turtles connect llm default deepseek
  ✓ Default LLM set to 'deepseek'.

The first server you add is automatically set as default.

turtles connect llm remove

$ turtles connect llm remove openai
  ✓ Removed LLM server 'openai'.

Using Ollama (Local LLM)

Run Turtles with zero API cost using a local model:

$ turtles connect llm add local

  🔌 Add LLM Server: local

  Select provider:
    1) OpenAI
    2) Anthropic
    3) Ollama (local)
    4) Custom (OpenAI-compatible)

  Choice [1-4]: 3

  Endpoint URL: http://localhost:11434/v1
  Model [llama4]: llama4

  ✓ LLM server 'local' added!
    Provider: ollama
    Model:    llama4
    Default:  no

  ℹ Test with: turtles connect llm test local

Set it as default to compile all Turtles locally:

$ turtles connect llm default local
  ✓ Default LLM set to 'local'.

Configuration

LLM configs are stored at ~/.turtles/llm/servers.json:

{
  "openai": {
    "provider": "openai",
    "endpoint": "https://api.openai.com/v1",
    "api_key": "sk-...",
    "model": "gpt-5.4"
  }
}

The default server name is stored at ~/.turtles/llm/default.