AI Agent Features
Configure optional AI model and web-search behavior.
Aripa includes configuration for AI-backed agent features and grounded web search. Only enable provider keys and model settings that you plan to use.
Model Providers
The configuration supports model selections for:
- Agent responses.
- Context summarization.
- Web search.
Provider keys are configured with environment variables:
| Variable | Used for |
|---|---|
OPENAI_API_KEY | OpenAI agent and summarizer models. |
OPENROUTER_API_KEY | OpenRouter model selections. |
AI_GATEWAY_API_KEY | Vercel AI Gateway model selections. |
GOOGLE_GENERATIVE_AI_API_KEY | Google Gemini grounded web search. |
Web Search
Web search is controlled by models.web:
{
"models": {
"web": {
"enabled": true,
"provider": "google",
"model": "gemini-2.5-flash"
}
}
}Set enabled to false to disable web search.
Rate Limits
Use agentRateLimitMessagesPerMinute to limit agent mentions per user:
{
"agentRateLimitMessagesPerMinute": 10
}Use null to disable the limit:
{
"agentRateLimitMessagesPerMinute": null
}Privacy and Cost
AI features may send Discord context, prompts, and tool outputs to the model providers you configure. Review each provider's privacy and retention policies before enabling these features.
Set logPrivacy to true if logs may leave the host machine:
{
"logPrivacy": true
}Model providers may charge for usage. Keep API keys private and monitor usage from the provider dashboard.