What is a Custom Integration?
A custom integration on Lutra is a collection of Python data models (dataclasses) and functions. In most cases, Lutra will generate the entire integration's Python code. It will also take care of making that code usable in a conversational AI context.
Key Concepts:
📦 A custom integration is just a Python file with:
Dataclasses (to define inputs and outputs)
Functions (to perform logic)
🚀 Any public function or dataclass is automatically exposed to Lutra's chat and playbook system. You can:
Call functions from chat
Chain them together in a playbook
Pass structured inputs and receive structured results
🙈 Private functions (those starting with an underscore, like
_helper_func) are ignored and won't show up in the chat experience.
🧠 What Makes Lutra's Plugin System Unique?
Lutra's custom integration system is batteries-included:
🔐 Secure Secrets & Auth:
Use Lutra's built-in authentication system to securely manage API keys, tokens, or credentials.
📝 Function Decorators:
Add descriptions and input metadata with simple decorators.
🐞 Debugging Utilities:
See what your function did at runtime—inputs, outputs, logs.
This makes every plugin easy to build, trust, and share.
🛠️ Plugin Auto Generation
You don't need to write all your code from scratch. Lutra supports conversational plugin development - ask Lutra to help you:
Create a new plugin based on documentation
Add or modify a function
Fix bugs
Add docs or error handling
🔒 Privacy & Sharing
By default: all plugins you create are private and only visible to your account.
On Lutra’s team plans, you can create plugins shared across your team.
If you're interested in making your custom integration publicly available to all users, please reach out to [email protected] and we'd love to be in touch on making it available to the broader Lutra community.