From the source
We're launching supabase.sh , a public SSH server that gives AI coding agents direct bash access to Supabase documentation.
It's free and available now as a public experiment Coding agents don't know what they don't know.
Without access to the latest docs, they guess - and often recommend non-existent APIs, run incorrect CLI commands, or skip steps like enabling RLS.
What's the best way to get up-to-date docs in front of them?
If you watch a coding agent work through a codebase, you'll see the same tools over and over: grep , find , ls , cat .
These models are trained heavily on bash and they treat the file system as their primary interface for exploration. supabase.sh gives them that same interface for our documentation.
Every doc page is a markdown file in a virtual file system at /supabase/docs .
Agents can search across topics, list pages, and read any page as needed.
The same workflow they use for code applies to docs.
Setting up your agent # Add recommended Supabase instructions to your project's AGENTS.md or CLAUDE.md : Or let your agent set it up for you: …
