FAQ

Common questions about Trodo prompt management: shipping without a deploy, labels vs versions, compile errors, caching, tools, and migrating existing prompts.

SDK reference

Every prompt-management method in the Node and Python SDKs — signatures, options, return shapes, and errors — plus the ManagedPrompt and CompiledPrompt fields.

Playground

An interactive bench to test one input across models side by side, score the outputs, and tune prompts — including specific managed-prompt versions — before you ship them.

FAQ

How do I ship a prompt change without redeploying my app?

What's the difference between a label and a tag?

What does compile() return — a string or messages?

Why did compile() throw before I even called the model?

A variable renders as literal {{name}} in my output. Why?

How do I pass conversation history to a prompt?

What happens if Trodo is unreachable when my app fetches a prompt?

How fast does moving a label propagate to my running app?

Can I pin an exact version instead of following a label?

Can a prompt call tools / MCP servers?

Does saving the same content twice create a duplicate version?

How do I A/B test two prompt versions?

Do the Node and Python SDKs behave the same?