Building Agents
Building Agents
Learn how to configure, customize, and run AI agents with supyagent.
Building Agents
Agents are the core unit of supyagent. Each agent is a YAML file that defines what model to use, what tools it can access, how it behaves, and who it can delegate to.
Agent Lifecycle
- Define — Write a YAML config in
agents/ - Chat — Interactive conversation with
supyagent chat - Run — Stateless execution with
supyagent run - Delegate — Connect agents together for complex workflows
Quick Example
name: coder
description: Expert programmer that builds, edits, and tests code
type: interactive
model:
provider: google/gemini-3-flash-preview
temperature: 0.3
max_tokens: 8192
system_prompt: |
You are an expert programmer. You write clean, tested, well-documented code.
When given a coding task:
1. Read existing code to understand the codebase
2. Plan your changes
3. Implement with small, focused edits
4. Test your changes
5. Report what you did
tools:
allow:
- "*"
will_create_tools: truesupyagent chat coderIn This Section
- Configuration — Full YAML schema reference
- Models — LiteLLM multi-provider setup
- System Prompts — Writing effective prompts
- Tool Permissions — Allow/deny patterns
- Sessions — Persistence and context management
- Chat Commands — Interactive mode commands