Basic Tutorial
The Basic tutorial guides you from zero to a production-ready agent in progressive steps. Each step builds on the previous one, and every step works on the same agent: a chatbot you scaffold in step 1, which gains configuration, policies, an HTTP tool and MCP tools along the way, and is finally built into an image.
This tutorial uses LangGraph, the recommended framework for building stateful agents. AgentVisor™ also supports interactive exec mode for running shells and AI coding tools. See Supported Frameworks for details.
Learning Path
| Step | Title | What You'll Learn |
|---|---|---|
| 1 | Your First Agent | Create the chatbot-agent project, run it, understand the code |
| 2 | Configuring Your Agent | Environment variables with .env files |
| 3 | Debugging Your Agent | Logging, log levels, access logs |
| 4 | Configuring AgentVisor | Runtime config file, environment variables, precedence |
| 5 | Adding Policies | Give the agent an HTTP tool; allow and deny hosts with MPE policies |
| 6 | Securing Your Agent's API | TLS encryption, identity, OIDC (OpenID Connect) |
| 7 | Connecting External Tools | Give the agent MCP tools, discovered and policy-checked |
| 8 | Deploying Your Agent | Build the agent into an image, deploy to production |
Prerequisites
Before starting, ensure you have:
- Docker installed and running
- AgentVisor CLI installed (see Getting Started)
- An AgentVisor license key (see Get a License Key)
jqinstalled (used throughout to pretty-print API responses in curl examples)
Skipping Ahead
Want the finished agent without the walk-through? agentvisor template create langgraph/tutorial-agent scaffolds the project as it stands at the end of step 7: the chatbot, the fetch_url tool, the filesystem MCP server, and the policies that govern them.
Next Steps
Start with Your First Agent to create and run an AI chatbot, then continue through the learning path to understand configuration, debugging, policies, and production deployment.