AI agents

This project ships three things for coding agents: an installable skill, plain-text mirrors of this site, and in-repo agent instructions. All three describe the same package — pick the one your setup consumes.

The skill

npx skills add oumarbarry/better-auth-py

That installs the better-auth-server skill — a SKILL.md plus four reference files — for Claude Code and any harness that reads the agent skills format (the CLI lists the supported ones). It teaches an agent to stand up a working server (FastAPI, Litestar, Flask or Django), protect routes, configure any of the 26 plugins and 35 social providers, migrate a Node Better Auth server onto the same database, and avoid the classic mistakes (short secrets, the in-memory default adapter, missing trusted_origins). Every code snippet in the skill has been executed and verified against the released package.

The skill covers using better-auth-server in your application. It is not a contribution guide — that is AGENTS.md below.

llms.txt

The site publishes both llmstxt.org endpoints:

Endpoint	Contents
/llms.txt	An index — every page with its URL and a one-line description
/llms-full.txt	Every page of this site, concatenated as one Markdown file

Fetch /llms.txt when the agent should pick the pages it needs; fetch /llms-full.txt when you want the whole documentation in context in one request. Both are regenerated on every deploy, so they never lag the site.

In-repo

AGENTS.md at the repository root governs agents contributing to the port: the parity prime directive (the TypeScript repo is canonical for anything touching wire or storage), the full test-and-lint gate to run before claiming work done, and the repo's conventions. CLAUDE.md points at it, so Claude Code picks it up automatically.

In short: install the skill to build with better-auth-server; read AGENTS.md to work on it.