Overview
Dexto's power comes from its customizability. You can customize every part of your Dexto agent with one yml config file.
Complete agent.yml
Complete reference for all agent.yml configuration options.
LLM Configuration
Configure the language model provider and settings for your Dexto agent.
System Prompt
Configure how your Dexto agent behaves and responds through system prompts that define personality, capabilities, and guidelines.
Memory
Configure the memory system to store and retrieve persistent information about user preferences, context, and important facts across conversations.
MCP Configuration
Configure Model Context Protocol (MCP) servers to extend your agent's capabilities by connecting to external tools, services, and APIs.
Permissions
Control how and when users are prompted to approve tool execution through Dexto's approval system.
Storage Configuration
Configure the storage input passed to the active image. The image resolves this input into
Sessions
Configure session management for your Dexto agent, including maximum concurrent sessions and session timeouts.
Telemetry
Enable distributed tracing to monitor agent behavior, debug issues, and track performance using OpenTelemetry.
Hooks Configuration
Extend agent behavior with custom logic that runs at specific points in the request/response lifecycle.
Tools
Tools are configured via the tools: list (tool factories). Each entry selects a factory by
Agent Card (A2A)
Configure your agent's public metadata for Agent-to-Agent (A2A) communication and service discovery.
Resources
What are Resources?
Dynamic Changes
Configure and manage runtime changes to agent state through the AgentStateManager.