Guides
Each guide starts from a problem you can hit in production with an AI SDK agent: why it happens, what fixing it by hand takes, and the change that fixes it with GNL.
Start with the core. Take the rest when you need it.
GNL is a full agent framework, durable from the core. You can use all of it, or put only the core into an agent you already have — and go from one to the other a step at a time.
Read the guide →Your AI SDK tool ran twice. Here's how to make it run once.
A retry, a redeploy or the model itself can run a side-effecting tool a second time. This page shows why, what fixing it by hand misses, and the change that stops it.
Read the guide →Your agent crashed halfway. Resume it where it stopped.
A restart should continue from the first step that never finished — not run the model and the finished tools again. This page shows why generateText cannot, what checkpointing by hand misses, and the change that does it.
Read the guide →Ask a human before your agent runs a tool. Keep the answer when the process dies.
Asking is the easy part. The hard part is the hours between the question and the answer, and the moment right after a yes.
Read the guide →