Skip to content

⚙️ Core Engine & Algorithmic Caches Reference

Technical specifications for Cortex Workstation's algorithmic caching data structures and engine coordination subsystems.

⚡ High-Throughput Eviction Caches

1. S3FifoCache

  • Module: cortex_unified.system_tools.s3_fifo
  • Architecture: State-of-the-art Simple, Scalable, Small (S3-FIFO) cache with small FIFO ($S$), main FIFO ($M$), and ghost FIFO ($G$) queues.
  • Key Properties:
  • Drastically reduces cache pollution from "one-hit wonder" scan operations.
  • Achieves higher hit-ratios than traditional LRU while requiring zero mutex lock contention during read hits.
  • Methods:
  • get(key: str) -> Any | None
  • put(key: str, value: Any)
  • stats() -> CacheStats

2. SieveCache

  • Module: cortex_unified.system_tools.sieve_cache
  • Architecture: SIEVE cache eviction algorithm using a circular queue hand pointer with visited bits.
  • Key Properties:
  • Ultra-lightweight with $O(1)$ amortized eviction time.
  • Ideal for rapid metadata indexing in directory walkers.

🚀 Orchestration Engine

  • Module: cortex_unified.core.engine
  • Class: Engine
  • Key Methods:
  • scan(categories: list[str]) -> ScanReport: Executes asynchronous multi-threaded discovery across all enabled system cleaner plugins.
  • clean(report: ScanReport, options: CleanOptions) -> CleanReport: Dispatches safe deletion tasks with automatic manifest recording and rollback checkpoints.