Short answer: Authored requests are maintained by hand and reflect what someone saved; generated requests are derived from the contract and reflect the current API. The first is a snapshot; the second is a build output. For a changing API, generation is the only way to stay correct without manual re-sync.
What authored requests guarantee
Exactly the calls someone built, with precise control. They do not guarantee currency, and they miss anything that was never saved.
What generated requests guarantee
| Property | Authored | Generated |
|---|---|---|
| Coverage | What was saved | Declared operations |
| On API change | Manual update | On save |
| Consistency | As maintained | With the contract |
| Control | Fine | Derived |
The failure that generation removes
A missing request for a new endpoint, or a stale request for a moved one, is the classic authored-set bug. A generated set reflects the spec, so it cannot be missing a declared operation.
The honest takeaway
Authored requests are the right tool for a small, stable, bespoke set. Once the API grows and changes, generating from the contract is the lower-maintenance, more correct path.
Return to the Powerduck vs Insomnia comparison, or read the API debugging guide.
Top comments (0)