Side quest

Prompt Lab

Run one prompt — or A vs B — across the same adapters the GEO audit uses. Live keys when present, labeled demo when not. Compare length, whether claims differ, and a stub preference rubric. This is not the audit/monitor/improve loop, and it will not help you plant false claims about third parties.

We optimize real public presence. We don’t fake consensus or plant false claims about third parties. Correction work updates real public sources. Trust & limits

Goal (optional)

What you want the models to do. Expand turns this into Prompt A and B.

Conductor (optional)

How to split the goal into variants — tone, length, constraints.

Prompt A *

Required unless you expand a goal. Same text is sent to every selected provider.

Prompt B (optional)

Leave empty to run A only. A/B comparison appears when B differs from A.

Providers

Same registry as audits. Live badge means a key is present on this server.

No run yet. Load the sample or paste Prompt A, pick providers, and run. We optimize real public presence — we don’t fake consensus.