Show HN: Sulci – a semantic cache that knows when not to answer

Category: ai-ml

Tags: llm-caching, semantic-cache, langchain, llamaindex, vector-search

Score: 6.8/10 (Innovation: 6, Technical: 7, Documentation: 7, Utility: 7)

Sulci is a semantic caching library for LLM applications that reuses responses for semantically similar queries and includes context-aware lookup for multi-turn conversations. It supports multiple vector backends, LangChain/LlamaIndex integrations, an MCP server, and a caching proxy, with benchmark harnesses for validation. The project stands out for its unusually candid documentation of retired claims and honest caveats about benchmarking.

Target audience: backend devs, AI engineers, devops

Repository: https://github.com/sulci-io/sulci-oss · Python · NOASSERTION · 11 stars

View on Hacker News