Swarm Learning Systems

Acerca de

Swarm Learning Systems is a specialized module for the OpenCode swarm plugin that enables autonomous improvement of multi-agent task decomposition. It monitors task duration, error rates, and success signals to calculate implicit feedback, ensuring that the most effective strategies are promoted while failing patterns are deprecated or inverted into anti-patterns. By applying confidence decay to stale data and providing a robust error accumulation system, it ensures that your agent workflows become smarter and more reliable with every execution.

Características Principales

• 359 GitHub stars
• Integrated error accumulator for detailed retry context and scoring
• Automated pattern maturity lifecycle from Candidate to Proven status
• Implicit feedback scoring based on duration, retries, and error signals
• Anti-pattern inversion for strategies with high failure rates
• Confidence decay logic to prevent dominance of stale performance data

Casos de Uso

• Debugging agent performance issues by tracking pattern effectiveness and error types
• Automatically identifying and avoiding ineffective task decomposition strategies
• Optimizing complex multi-agent coding workflows through historical analysis