Show HN: TS Compiler Knowledge Graph reducing AI tokens about 90% (github.com)

The introduction of the @ttsc/graph MCP server is a game-changer for AI agents working with TypeScript codebases, reducing AI token usage by approximately 90%. This server creates a precise code graph from TypeScript files, indexing declarations and their relationships rather than relying on traditional file reading methods. The conventional approach necessitates extensive file searches, leading to high token costs in AI interactions. By utilizing the TypeScript compiler to generate this graph, agents can obtain answers through efficient single calls without needing to traverse files, resulting in significantly lower computational expenses. This innovation is particularly significant for the AI/ML community as it streamlines agents' interactions with code, fostering more efficient and reliable communication. Unlike existing tools like codegraph and serena that experience fluctuating costs based on repository size, @ttsc/graph maintains consistent token usage. With a graph that builds rapidly—taking just 29 seconds for large repositories—this tool enhances the developer experience by enabling quick access to architecture, runtime flows, and symbol details. Such advancements not only facilitate a more accurate understanding of codebases but also pave the way for sophisticated AI-driven code analyses and interactions.