SoL-Pi: Scaling Auto-Research Loops for Efficient Agent Harnesses

Unified algorithmic compaction + observational memory for Pi. Merges pi-vcc and pi-observational-memory with fallback chains, cooldowns, and per-worker model config.

稳定可用 Use billion-context instead .Active Context Pruning for pi ，100K tokens is enough. ，主动压缩上下文，10w 上下文足矣

A well-commented, ready-to-use Pi Agent configuration starter that helps you actually understand what's happening.