📄️ SQLite Dashboard Caching
Cache read-heavy SQLite queries for a live dashboard. Bust specific queries when data changes without flushing everything.
📄️ AI Skill System with Hot-Reload
Load hundreds of AI skill prompts from markdown files and cache them. Edit a file and the cache invalidates automatically. No server restart needed.
📄️ Multi-Layer RAG Pipeline Caching
Cache vector search results, prompt compilation, and LLM responses in a three-tier RAG pipeline. Use context versioning to invalidate everything when your knowledge base updates.
📄️ Surgical Cache Invalidation
Evict one cached entry at a time with cache_invalidate(). Drop session and request objects from cache keys using exclude=(). No more blunt cache_clear().
📄️ Cache Observability with Store Stats
Monitor hit ratios, eviction rates, and cache sizes with MemoryStore.stats(). Export to Prometheus, set up health checks, and debug cache effectiveness in production.
📄️ Production API with HTTP Caching
Build a production FastAPI endpoint with in-process caching, Cache-Control headers, ETag/304 support, parameter exclusion, and surgical invalidation. Full stack from v1.0.
📄️ User Session Caching with Sliding TTL
Keep active user sessions warm in the cache with sliding TTL. Sessions expire naturally when users go idle, without arbitrary mid-activity logouts.
📄️ Cache Warmup Strategies
Pre-populate the cache at application startup to eliminate cold-start latency. Warm multiple stores, handle partial failures, and monitor warmup progress.
📄️ Dynamic TTL with Feature Flags
Change cache TTL at runtime using callable TTLs driven by feature flags, environment variables, or remote config. No redeploy needed.
📄️ Framework-Agnostic HTTP Caching
Use inhouse's core HTTP cache primitives with Flask, Django, or any Python web framework. ETag matching, Cache-Control headers, and 304 responses without FastAPI.
📄️ Multi-Tier Caching with Separate Stores
Run multiple MemoryStore instances with different TTLs, capacities, and eviction policies. Isolate hot user data from cold reference data for optimal cache utilization.