Deploying your first Agent
Deploy a custom container as a Krypton Agent.
Tutorials
Guided paths for deploying agents, MCP servers, and self-hosted LLMs.
These guides are task-oriented. Each one starts from a working Krypton install, applies a concrete resource, verifies the endpoint, and points to the reference page for deeper configuration.
Choose a path
| Goal | Start here |
|---|---|
| Deploy an A2A, plain HTTP, ADK, or LangGraph container | Deploy your first Agent |
| Host a GGUF model from Hugging Face and call it with an OpenAI SDK | Deploy your first LLM |
| Run an MCP server over HTTP or wrap a stdio MCP binary | Deploy your first MCP server |
Before you begin
Install Krypton first:
helm install krypton oci://ghcr.io/kryptonhq/charts/krypton \
--namespace krypton-system --create-namespace
For a laptop loop with kind, use Local testing.
Deploy Your First LLM
Host a Hugging Face model and call it through the OpenAI API.
Deploying your first MCP server
Host any Model Context Protocol server as a Krypton Agent.