# XUSS — llms.txt

> XUSS (https://xuss.us) — EU NVMe hosting with full SSH, AI coding agents, 24/7
> Python/Node services and a scoped Hosting API. This file lists the machine-readable entry points.

## Docs
- [Complete guide for AI agents](https://xuss.us/agents.md): everything — plans (incl. a free plan), payments, SSH, the six pre-installed AI agents, the built-in panel assistant, `run`, databases, backups, subdomains, the community chat, the OpenAI-compatible AI API, the scoped Hosting API, support.
- [Hosting API (agent guide)](https://xuss.us/agents.md#9c-hosting-api-scoped-control-for-automation): scoped `xsk-` hosting tokens (full/read-only/custom permissions), direct lifecycle endpoints, panel resource proxy, safety rules and examples. Human token management/docs: https://xuss.us/panel/hosting-api.
- [AI API docs](https://xuss.us/ai-api.md): OpenAI-compatible API at https://xuss.us/v1 (endpoints, auth, streaming, models).
- [Knowledge Base index](https://xuss.us/api/articles?lang=en): JSON list of all articles (langs: en, ru, uz).
- [Plans JSON](https://xuss.us/api/plans): live plans, prices, limits.
- [Sitemap](https://xuss.us/sitemap.xml)

## Notes
- /panel/* and /admin/* require login (excluded in robots.txt).
- Login/registration are captcha-protected; Hosting API token creation is a human panel action.
- Hosting API tokens are separate from AI API keys: `scope=hosting` cannot call `/v1`, and `scope=ai` cannot call `/api/v1/hosting`. Hosting tokens can be full-access, read-only, or custom-capability.
- AI agents need RAM: opencode ~1 GB per instance (Pro+ recommended); hermes ~240 MB fits Basic.