Skip to main content

Module smart_retrieval

offline_intelligence::context_engine

Module smart_retrieval 

Source
Expand description

Smart retrieval with hierarchical context optimization

This module implements intelligent retrieval that minimizes recomputation cost by preferring summaries over raw messages and enforcing strict token budgets.

Key optimizations:

  • Tier 1 (hot cache) → O(1) return
  • Tier 2 (summaries) → Compressed context (50 tokens vs 2000)
  • Tier 3 (database) → Importance-filtered, token-budgeted
  • Hierarchical assembly → Summary first, details on-demand

Structs§

RetrievalResult
Result of smart retrieval operation
SmartRetrieval
Smart retrieval orchestrator
SmartRetrievalConfig
Configuration for smart retrieval

Enums§

RetrievalStrategy
Strategy used for retrieval