I cut my agent token usage in half (github.com)

indexio, a new code search engine designed for AI coding agents, has been launched, offering significant improvements in performance and efficiency. Built as a single binary in Rust, indexio eliminates the need for databases or external services by indexing an organization's repositories into a straightforward file directory. This indexing process allows agents to respond to queries in milliseconds via the Model Context Protocol (MCP) and HTTP API, effectively halving the token usage for searches by harnessing its own indexing and search capabilities. This tool is particularly noteworthy for the AI/ML community as it enables enhanced code searching and indexing without extensive infrastructure. indexio allows users to index git repositories and other code sources seamlessly, search with hybrid queries that combine lexical and semantic methods, and continuously sync indexed data with minimal overhead. The implementation can be easily deployed across various platforms, and its innovative architecture ensures that it learns from the specific vocabulary of an organization, providing tailored results. This could revolutionize how developers interact with their codebases, making it faster and more efficient to locate relevant information and navigate complex projects.