Show HN: LLMSchema – Dump your database schema as compact Markdown for AI agents

LLMSchema is a new open-source tool that converts database schemas into compact Markdown documentation designed for AI agents and coding assistants. The developer shared the project on Hacker News as a Show HN post, releasing it under an MIT license.

The tool works as a single Go binary and can function either as a CLI or a library. According to the GitHub repository, it supports Postgres, MySQL, and SQLite databases. The creator built LLMSchema after observing that AI coding agents would inefficiently infer database schemas by reading migration files and application code. By generating compact schema documentation in advance, agents can reference a centralized source of truth instead of extracting schema information from scattered source files.

The developer maintains the schema documentation by running LLMSchema as part of their migration command, keeping it automatically synchronized with database changes. The Markdown output can be referenced directly in agent prompts or included in project documentation files, reducing the need for agents to reconstruct schema information during each task. This approach prioritizes efficiency in how AI systems understand and interact with databases.