AskAI
This plugin lets you chat with your favourite AI models in-game. Supports OpenAI, Anthropic, and Google Gemini.
On 'player' mode, each player picks a provider and model, sends messages with /chat, and gets private AI responses. On 'server' mode, the server sets a single API key and model for all players to use freely.
Note
This plugin is not meant for long conversations. It operates as a single shot Q&A format. Each message is sent with fresh context.
Requirements
- Paper or Purpur 1.20.6+
Commands
| Command | Description |
|---|---|
/chat <message> |
Send a message to your active AI provider |
/chat setkey <provider> <key> |
Set your API key for a provider |
/chat setmodel <provider> <model> |
Set which model to use for a provider |
/chat provider <provider> |
Switch your active provider |
/chat status |
Show your current config and key status |
/chat reload |
Reload config.yml without restarting |
Providers: openai, anthropic, gemini
Permissions
| Permission | Default | Description |
|---|---|---|
askai.use |
All players | Use /chat and /chat status |
askai.setkey |
All players | Manage keys, models, and provider selection |
askai.admin |
OP only | Set server-wide API keys in server-key mode, reload config |
Key Modes
Player mode (default)
Each player sets and manages their own API keys. This is the default behavior.
Server mode
One set of API keys is shared by all players, managed by admins. Players can still choose their own provider and model, but the API key comes from the server.
To enable, set key-mode: "server" in config.yml and run '/chat reload'. Only players with askai.admin permission can set keys in this mode.
Configuration
# Encryption seed for API key storage. Set this once before first use.
# Changing it after players have set keys makes their keys unrecoverable.
encryption:
seed: "CHANGE-ME-use-a-long-random-string-here"
# "player" (each player sets own keys) or "server" (admin sets shared keys)
key-mode: "player"
# Rate limiting per player
rate-limit:
requests: 10
window-seconds: 60
# Message limits
max-message-length: 2000
max-response-tokens: 1024
# System prompt prepended to all conversations
system-prompt: "You are a helpful assistant in a Minecraft server."
# Which providers are enabled
allowed-providers:
- openai
- anthropic
- gemini
