Show HN: Amdb – Tree-sitter based memory for AI agents (Rust) (github.com)

A new tool called Amdb (Agent Memory Database) has been introduced to enhance AI coding assistants by providing them with a deeper structural understanding of entire codebases. Unlike existing AI models that are limited to the files users open, Amdb scans local projects, builds a vector index of the code, and generates a comprehensive Markdown context file. This allows AI models to better grasp the nuances of projects by leveraging detailed insights derived from the codebase. The tool is user-friendly, available for macOS and Linux, and requires no prior Rust installation, enabling easy implementation through a simple script or installation command. The significance of Amdb lies in its ability to transform the way developers interact with AI assistants by providing them with “God Mode” access to project semantics. Utilizing robust Tree-sitter parsers for various programming languages—from Rust and Python to Java and C#—Amdb captures the syntax and structure of the code, allowing for targeted context generation. This feature not only streamlines the coding process, particularly for large projects, but also helps focus on specific functionalities, such as authentication logic. By integrating this advanced memory solution, developers can unlock a more productive and informed collaboration with AI, ultimately leading to better coding outcomes.