Souveraineté

Installation & First Steps

Souveraineté works out of the box, but for full immersion, you need to configure the AI ​​provider.

1. Make sure your server have PacketEvents plugin.

2. (Optional) Install Villages Revamped datapack.

3. Download Souveraineté, place the .jar file in your server’s plugins folder, and start the server.

4. Stop the server and go to the plugins directory.

5. Find the provider.yml file. This is the file you need to configure.

provider.yml

Setting up a provider is easy. All you need to do is select a service and generate an API key. Some services offer free trials, so you don't have to pay to use their AI. For free use, I recommend Groq: it's fast and not too restrictive. If you're considering paid services, I recommend DeepSeek: it offers the lowest prices on the market and good quality.

# ProviderConfiguration Configuration File

# The provider type for content generation. Choose between GEMINI, OPENROUTER, GROQ, DEEPSEEK, CHATGPT, ANYTHINGLLM.
# GEMINI (aistudio.google.com) have a free tier, but since December 7 (2025), it's unusable because of too low request limitation (20 RPD).
# OPENROUTER (openrouter.ai) requires payment (with competitive pricing; however trial is available). The best option, but paid.
# GROQ (console.groq.com) is free and recommended. Suitable for testing and gaming, but may be too limited for production (if we are talking about a free plan and large servers).
# DEEPSEEK (deepseek.com) is paid-only but the cheapest option.
# CHATGPT (chatgpt.com) - The standard OpenAI API.
# ANYTHINGLLM (useanything.com) - Self-hosted solution. Requires running the software locally or on a server. Fully private and free (depending on your local backend like Ollama).
provider-type: GROQ

# For GEMINI, default model is "gemini-2.5-flash-lite".
# For OPENROUTER, default model is "google/gemini-2.5-flash-lite".
# For GROQ, I recommend "openai/gpt-oss-120b".
# For DEEPSEEK, default model is "deepseek-chat".
# For ANYTHINGLLM, this usually depends on your workspace settings. You can often leave it as "gpt-3.5-turbo" for compatibility.
model: openai/gpt-oss-120b

# The API Endpoint URL. Required ONLY for self-hosted providers like ANYTHINGLLM.
# Standard local address for AnythingLLM is "http://localhost:3001".
# If you are using cloud providers (Gemini, Groq, OpenRouter, etc.), this field is ignored.
url: http://localhost:3001

# Paste your API key here.
api-key: YOUR_API_KEY

# The language for generated content. Specify the desired language (e.g., 'English', 'Spanish', 'Russian', 'Dalek Language', 'Moonspeak', etc.).
language: English

# The naming convention for generated content. For example, 'English Names' for standard English-style names.
naming-style: Fantasy Names

# The thematic setting for content generation. For example, 'Fantasy' for a fantasy-themed world.
setting: Fantasy

# Controls the randomness of generated content. Higher values (e.g., 2.0) increase creativity but may reduce coherence.
temperature: 2.0

# The maximum number of retry attempts after a failed content generation request.
max-retries: 1

# Proxy configuration for connecting to Gemini in regions where it is restricted (e.g., Russia, China). Won't be used if host name is PROXY_HOST.
proxy: 
  # The type of proxy to use (e.g., HTTP, SOCKS).
  type: HTTP
  # The proxy server hostname or IP address.
  host: PROXY_HOST
  # The port number for the proxy server.
  port: 1337
  # The username for proxy authentication, if required.
  user: PROXY_USERNAME
  # The password for proxy authentication, if required.
  pass: PROXY_PASSWORD