Show HN: Mantic.sh – Search 480k files in 0.46s without embeddings (github.com)

Mantic.sh has unveiled a cutting-edge code search engine that dramatically enhances the speed and efficiency of searching through large codebases, managing to rank files in under 500 milliseconds without relying on embeddings, vector databases, or external dependencies. This innovation allows AI agents to quickly infer user intent from metadata and file structure instead of scanning entire file contents, significantly reducing processing time and token usage—by as much as 63%. The tool also prioritizes privacy, operating entirely on local systems with no data egress. The release addresses critical issues from previous versions, such as improved scoring for all queries and enhanced reliability by resolving known operational errors. Mantic’s architecture includes features like deterministically scoring results, impact analysis for assessing changes, and native integration with popular AI tools like Claude Desktop. The potential for cost savings is considerable, positioning Mantic as a competitive alternative to cloud-based solutions, with zero annual costs for teams compared to tens of thousands for existing SaaS models. This development is poised to benefit developers by streamlining their workflow and enhancing code retrieval capabilities significantly.