Agent Settings
The agents.defaults section controls the core agent behavior.
{
"agents": {
"defaults": {
"workspace": "~/.kyber/workspace",
"model": "google/gemini-3-flash-preview",
"provider": "openrouter",
"maxTokens": 8192,
"temperature": 0.7,
"maxToolIterations": 20
}
}
}Fields
| Field | Type | Default | Description |
|---|---|---|---|
workspace | string | ~/.kyber/workspace | Path to the agent’s workspace directory |
model | string | google/gemini-3-flash-preview | Model identifier (provider-specific) |
provider | string | openrouter | Which provider to use (see Providers) |
maxTokens | number | 8192 | Maximum tokens in the model response |
temperature | number | 0.7 | Sampling temperature (0.0 = deterministic, 1.0 = creative) |
maxToolIterations | number | 20 | Maximum number of tool calls per message before the agent stops |
Model selection
The model field uses the provider’s model identifier. Examples:
| Provider | Example model |
|---|---|
| OpenRouter | anthropic/claude-sonnet-4-20250514, google/gemini-3-flash-preview |
| Anthropic | claude-sonnet-4-20250514 |
| OpenAI | gpt-4o |
| Gemini | gemini-2.5-flash |
| DeepSeek | deepseek-chat |
| Groq | llama-3.3-70b-versatile |
Tool iterations
The maxToolIterations limit prevents runaway tool loops. If the agent hits this limit, it stops calling tools and responds with what it has. Increase this for complex multi-step tasks.