Advanced concepts
In this section you will find concepts that apply to advanced use cases:
The Toolhouse Agent File (also called th file)
Learn how to use advanced configuration features with the Toolhouse Agent File.
Build agents with the th fileAgent preview and testing
Each Toolhouse agent comes with a preview environment that helps you test your changes before publishing it.
Test agents before publishingPublish and run your agents
This guide shows you how to publish your agents. Each agent can be used on its own or integrated into third party platforms.
Publish and run your agentsBuild stateful agents
Each agent is aware of its past interaction with each one of your users (or with you, if you're the only user of your agent), without the need to integrate databases or memory tools. Learn how to make the most out of this functionality.
Build stateful agentsBring your Model
Each agent uses the default Toolhouse model. Sometimes you want to switch model, for example if you want to use your own models, or if you need a more powerful model.
Bring your ModelManaging secrets
In order to use your models, you will need to add API Keys to Toolhouse. This page illustrates how to do it.
Managing secretsBundle MCP servers together
Toolhouse offers a way to group MCP servers together. This is useful to restrict the type of actions your agent can take.
Bundle MCP servers togetherCustomize agents for your end users
Your agent can understand which user it's talking to. Learn how to configure this behavior.
Customize agents for your end usersGenerate a Vibe Coding Prompt with the Toolhouse CLI
Learn how to use the Toolhouse Command Line Interface (CLI) to create a one-shot prompt for common services like Lovable, v0, or Cursor.
Generate a Vibe Coding Prompt with Toolhouse CLILast updated