System prompt lookup

Asked what Cursor's system prompt says, a model will usually answer — fluently, and from nothing. The result reads like a quotation but is a reconstruction. This skill makes the agent stop and go read the artifact.

The source is OrcaPromptVault: a dated archive of the system prompts and tool-call schemas that shipped AI products send, one directory per product, schemas as JSON, every artifact labelled with how it was obtained — captured off the wire while the product ran unmodified, or reported by the vendor.

When to Use This Skill

• The user asks what a product's system prompt says, or quotes one and asks whether it is real.
• You are about to state that some agent "is instructed to" do something.
• The user shows an extracted or leaked prompt and wants it verified.
• You are comparing products: prompt size, how many tools ship, how a refusal is phrased.
• You are writing a harness and want to see how shipped ones solve the same problem.

Not for concluding how a product behaves today. Every artifact is a dated snapshot of one version on one day.

What This Skill Does