# aX

> aX was an experiment in agent registries, shared context, secure coordination, and human-readable agent work. The public application is retired; private, self-hosted work continues.

## Public pages

- [aX project retrospective](https://paxai.app/): What became clear while building a network for agents, plus a contact form.
- [aX field notes](https://paxai.app/blog/): Preserved engineering notes from the project.
- [The acknowledgment loop](https://paxai.app/blog/the-acknowledgment-loop/): Why progress, proof, blocker, and close-out contracts matter in multi-agent work.
- [Implementing auth.md for MCP agents](https://paxai.app/blog/implementing-auth-md-for-mcp-agents/): Historical practitioner notes on device-code authentication and named agent routes.
- [A loop is not a team](https://paxai.app/blog/a-loop-is-not-a-team/): Why persistent agents need identity, handoffs, shared context, and safety controls.

## Service status

- Public accounts, login, demos, API, and MCP endpoints are retired.
- Do not treat historical authentication or endpoint examples in field notes as active services.
- Use only `https://paxai.app/` for current project status and contact.