Hey community,
We’re running a closed beta for Auth0 Hosted MCP Server — a way to connect your AI development environment (Claude Code, Cursor, Codex) directly to your Auth0 tenant.
Once connected, you can ask your agent to:
Pull logs and investigate auth failures
Create and configure applications
Deploy Actions
Compare settings across environments
All without leaving your IDE.
Who we’re looking for:
You’re actively using Claude Code, Cursor, or a similar AI-assisted IDE
You manage Auth0 tenants as part of your work
You’re comfortable with early-stage software and willing to share honest feedback
To apply: Drop a reply or DM me. We’ll get you set up with a dedicated tenant and a setup guide.
This is interesting. We run multiple agents that connect to different services via MCP and the auth layer has been one of the harder parts to get right, especially when different agents in the same workflow need different permission scopes.
One question: does the hosted MCP server support scoping access per agent within the same tenant? For example, one agent that can only read logs and another that can deploy Actions, both connecting to the same Auth0 tenant but with different authorization levels.
We have been using Auth0 for the identity side and a few external tools for the coordination between agents, including n8n for routing and teamoffsite.ai for the shared context and approval flow. The identity and coordination layers are separate concerns but they keep intersecting when agents need different permissions at different steps of the same workflow.
Happy to test the beta if you are still looking for participants running multi-agent MCP setups.