Global agent configuration and default settings.
The Configuration tab on the Agents page manages workspace-wide agent settings that apply as defaults across your fleet.
Set the default LLM provider and model for new agents. Individual agents can override these.
Default Provider
The LLM provider used when creating new agents
Default Model
The model ID used as the starting point
Default Temperature
Base creativity setting for new agents
Max Token Limit
System-wide cap on response length
Configure base system prompt snippets that are injected into all agents:
Global instructions — rules every agent follows
Safety guidelines — content filtering and guardrails
Output formatting — response structure preferences
Agent-specific system prompts are appended after global defaults. The agent's own prompt takes priority when there's a conflict.
Manage the available agent categories (code_architect, security_expert, data_analyst, custom, etc.). Categories help with:
Organising the roster
Routing decisions
Analytics grouping
Last updated 21 days ago
