Skip to content

What is Nyxory?

Nyxory is the runtime your AI coding agent uses to ship production software. Your agent calls our tools, applies our recipes, and gets a live URL back — no manifests written by hand, no dashboards to babysit.

Three entry points over MCP, plus a cookbook of recipes indexed by stack and topic:

  • nyx_deploy_start — scans the repo, returns the relevant recipe + your project’s current state, and lets the agent take it from there.
  • nyx_help — bidirectional. When your agent gets stuck on something specific, it asks back; we return the recipe content that addresses it.
  • The nyx_* primitives — the actual tools the agent uses to do work: create apps, set secrets, attach domains, check rollouts.

A live URL within minutes. The platform takes care of the parts agents are bad at: clusters, certs, ingress, scaling, the long tail of last-mile deployment. You stay in your coding agent the whole time.

  • Not a deploy state machine. We never override your agent’s reasoning. The recipe is a suggestion shaped by experience, not a fixed sequence.
  • Not an LLM that re-thinks what your agent already does well. We don’t generate plans at runtime; recipes are static, versioned, and authored by us.
  • Not lock-in. Your manifests, secrets, and data are exportable at any time.

Quickstart →