Skip to main content
Score ideas from a script or backend: viability score, competitors, source-linked evidence. One header, no SDK. For Claude, Cursor, or Codex, use Get started instead.
1

Get your key

Calling the API from your own code needs a key you hold yourself, so create one by hand at Account → API Keys, self-serve on Founder and above. That is the only path that shows you a prv_... value. Send it as x-preuve-key (see Authentication).
Copy the key when you create it. It’s shown once. Keep it out of git and out of browser code.
Connecting the Preuve connector instead works on a free account, but it mints the key inside your MCP client and never shows you the value, so there is nothing to send as x-preuve-key. For that path, start at Get started.
2

Create an analysis

scanType is required: "starter" is free, "deep" spends paid quota. clientRunId is how you retry: sending the same value again returns the existing run instead of starting a second one. Optional context fields sharpen the report: stage (idea | validation | mvp | launched | growth | scaling) and budget (bootstrap | 10k | 50k | 100k | 100kPlus | 1mPlus) feed the feasibility analysis directly — the MCP tools make both required on deep scans.
3

Poll until complete

Starter runs typically complete in 1-3 minutes, deep runs in 5-10. Poll every 10-15 seconds until status is COMPLETED.
4

Enrich, then export

You get stable JSON (schemaVersion: 2): score, verdict, competitors, risks, and citations. Deep scans also include the full report, pivots, and any modules. See Scans and enrichment.

Next steps

MCP server

Same verdict, from the chat you are already in.

Batches

Score up to 10 ideas and rank them by viability.

Deep modules

Proof of Demand, Founder Fit, Playbook, and Trends on demand.

Quotas and billing

What gets charged, and what gets refunded.