For AI Agents¶

Guidance specifically for AI agents working with LionAGI.

Agent Resources¶

• Orchestration Guide - How to orchestrate multi-agent workflows
• Pattern Selection - Choose the right patterns for your task
• Claude Code Usage - LionAGI in Claude Code environment
• Self-Improvement - Learn and adapt your approaches

Key Concepts for AI Agents¶

• Think in Graphs: LionAGI uses dependency graphs, not sequential chains
• Leverage Parallelism: Multiple operations can run simultaneously
• Use Builder Pattern: Construct workflows declaratively
• Handle Context: Each Branch maintains persistent memory

Common Agent Workflows¶

1. Research & Analysis: Fan-out for parallel research, fan-in for synthesis
2. Multi-Perspective: Different agents provide specialized viewpoints
3. Iterative Refinement: Sequential operations that build on each other
4. Validation Patterns: Multiple agents validate and critique results

Best Practices¶

• Start with simple patterns before complex orchestration
• Use appropriate concurrency levels for your resources
• Monitor costs and performance as you scale
• Handle errors gracefully with fallback strategies