What’s this?
Prompt Brain is a pre-flight companion for apps that call the Graydient Render API. You send the same request fields you would send to Graydient; Prompt Brain never starts a render. It returns tips, a cleaned suggested_prompt, runtime_hints, and will_likely_fail so you can fix the job first.
/v1/simulate are in beta for API customers. Shadow mode is on: treat guidance as advisory while we harden behavior.Ambition: make every render succeed — or retry intelligently — with no extra coding on your side. Drop in the same Graydient request shape; Prompt Brain aims to catch timeouts, missing media, bad workflows, VRAM blow-ups, and secret leaks before you spend a GPU-second.
- Same fields as Graydient Render (prompt, workflow slug, params, images) — drop-in pre-check.
- At most one automatic retry guidance; never loops; never queues a render.
- Built for partner apps and agent clients that already call Graydient.
Official production API (V3): Graydient AI rendering API (V3) · detailed reference: API Help.
Try a failure before you render
Pick a synthetic Example, inspect the editable request, run /v1/simulate/demo, and read why it fails and how Prompt Brain helps. Examples are never real customer prompts.
Live response
{}
Why it fails on Graydient
Select an example.
How Prompt Brain helps
Select an example.
From the live response
Run to see live fields.