Show HN: Veles – Hybrid (BM25 and semantic) local code search MCP, in Rust (github.com)

Veles, a fast hybrid local code search tool written in pure Rust, has been introduced to facilitate efficient querying for both AI agents and developers. It uniquely combines BM25 and semantic search methods, allowing for rapid query responses in tens of milliseconds, thanks to its entirely CPU-based architecture and advanced indexing techniques. Veles supports on-disk persistent indexing, which enables incremental updates while maintaining a cache of previously indexed data. Key features include integration capabilities through MCP and gRPC servers, tree-sitter symbol lookups, and multiple output formats to suit various development environments. This release is significant for the AI/ML community as it enhances local code search functionality, a vital aspect of code navigation and understanding in AI workflows. The hybrid search model employs Reciprocal Rank Fusion (RRF) for improved ranking accuracy, complemented by features like identifier-aware tokenization and the ability to filter search results across different programming languages. With prebuilt binaries for major operating systems and a rich set of APIs for integration with existing AI tools, Veles streamlines the search process, enabling developers to quickly locate and utilize code snippets, thereby accelerating development timelines and enhancing productivity.