multi-agent-patterns - Multi-agent architecture patterns for task distribution.
Provides architectural patterns to distribute work across language model instances with isolated context windows.
Tags
Updated: 2026-09-16Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Distribute work across multiple LMs
- Implement supervisor orchestrator pattern
- Implement peer to peer pattern
- Implement hierarchical agent pattern
- Isolate context windows across agents
- Pass messages between agents
- Execute weighted voting consensus
Inputs
- User queries
- Task specifications
- System prompts
- Tool definitions
Outputs
- Final response outputs
- Direct responses to user
- Inter-agent messages
- Distilled task summaries
- Persistent memory files
Requirements
- Multiple LM instances
- Persistent file system storage