dbdocs vs LLMSchema

dbdocs creates web-based database documentation from DBML definitions. DBML is an open-source markup language for defining database structures; published documentation lets readers browse tables and fields and view relationships between tables. Teams can store DBML locally or in Git, then publish it to a dbdocs project. The db2dbml command supports imports from PostgreSQL, MySQL, MSSQL, Snowflake, and BigQuery. Documentation views can be embedded in Notion, Confluence, Miro, or a website. CI publishing uses a token, and the instructions include a GitHub Actions example. New projects are public by default. The Free plan lists password-protected sharing, while private mode is available on paid plans. The cloud service includes schema history and a data dictionary, and its stated database coverage is relational. dbdocs is aimed at internal development teams and software development agencies. The Free plan is free. Personal Pro costs $20 a month, or $16/month annually; Team costs $60 a month, or $50/month annually.

LLMSchema is free, MIT-licensed software that generates database schema documentation for LLMs and AI agents. It extracts relational schemas from PostgreSQL, MySQL, and SQLite, then documents tables, columns, types, indexes, constraints, and relationships in Markdown. The default output is one portable Markdown document. An alternate layout creates an overview plus a separate Markdown file for each table. Users can include selected tables or exclude tables from the generated documentation. LLMSchema is available as a command-line tool and a Go library, with installation instructions for Go and a quick install script for macOS and Linux. The README shows adding schema generation to a Makefile or migration script to keep documentation current. It also suggests pointing a coding agent to the schema files from AGENTS.md, CLAUDE.md, or an equivalent instruction file. The project is self-hosted and listed for Linux, macOS, and Windows. Its maker positions it for development databases and says not to rely on it for production-critical documentation.