Validate the idea before your agent builds it.

Your agent ships in minutes, butonly 17.5% of ideas earn a go. Preuve runs the full, source-linked viability check first - from inside your agent.

API keys are tied to a paid Preuve account. Create yours, then generate the key here.

7 tools appear: start_analysis, get_analysis, enrich_analysis, ...

Paste any of these into claude.ai, Claude Code, or Cursor. Claude picks the right Preuve tool on its own.

Keys are self-serve for paid accounts and carry explicit scopes. Analysis usage is charged to the owning account quota - Starter scans at no cost, deep scans from subscription, lifetime, or token balance.

POST/api/agent/analysesanalysis:writeStart one analysis (starter or deep)

GET/api/agent/analyses/:idanalysis:readPoll one analysis

POST/api/agent/analyses/:id/enrichanalysis:writeGenerate export sections + deep modules

GET/api/agent/analyses/:id/exportexport:readExport one analysis as JSON

POST/api/agent/analysis-batchesbatch:writeStart up to 10 analyses

GET/api/agent/analysis-batches/:idbatch:readPoll a batch

GET/api/agent/analysis-batches/:id/exportexport:readExport a batch as JSON

One header:x-preuve-key, holding the whole key. No signing, no timestamps, no nonces.

curl https://preuve.ai/api/agent/analyses \ -H "x-preuve-key: $PREUVE_API_KEY" \ -H "content-type: application/json" \ --data '{"clientRunId":"run-online-001","scanType":"starter","idea":"A workflow automation service for solo founders that validates market demand before they build.","publish":true}'

KeepPREUVE_API_KEYserver-side. The key carries no origin restriction, so it never belongs in browser code.

async function preuve(path, { method = 'GET', body } = {}) { return fetch(https://preuve.ai${path}, { method, headers: { 'content-type': 'application/json', 'x-preuve-key': process.env.PREUVE_API_KEY, }, body: body ? JSON.stringify(body) : undefined, }); }

Scopes are exact. Write access for analyses does not grant export access.

analysis:writeCreate analysis runs and enrich them for the key owner

analysis:readPoll single-run status, module states, and URLs

batch:writeCreate bounded batches up to 10 ideas

batch:readPoll batch status and child run URLs

export:readExport completed analyses and batches as JSON

Polling returns status, enrichment progress, per-module states, and URLs. The export endpoint returns the structured report itself (ideas-json, schemaVersion 2) - full deep sections included on paid runs.

{ "id": "run_123", "reportId": "report_123", "clientRunId": "run-online-001", "scanType": "deep", "analysisTier": "advanced", "status": "COMPLETED", "readyForExport": true, "enrichment": { "mode": "core", "status": "completed" }, "modules": { "proofOfDemand": { "status": "completed" }, "founderFit": { "status": "not_generated" }, "playbook": { "status": "generating" }, "trends": { "status": "not_generated" } }, "reportUrl": "https://preuve.ai/app/results?id=report_123", "shareUrl": "https://preuve.ai/share/abc123" }

Paid deep reports can generate four optional modules through the enrich endpoint. One successful generation per module per report, included in the deep scan price - a failed generation never consumes the cap.

proofOfDemandVerified real-demand prospects with verbatim quotes. Beta: results can be uneven, every quote links to its source.async

founderFitFounder-vs-plan underwriting from a submitted founder profile: fit score, top risks, pivot fit.inline

playbookA concrete launch playbook generated for the validated idea.async

trendsGoogle Trends data for the idea, generated on demand - the analysis never fetches it on its own.async

Batches aggregate child runs intoPROCESSING,COMPLETED,PARTIAL, orFAILED.

PENDINGThe run record exists and is waiting for dispatch

PROCESSINGThe Preuve analysis pipeline is running

COMPLETEDThe report is done - enrich if needed, then export

FAILEDThe run reached a terminal error with a stable code

Structured export for single runs and batches

What a key can do - and what it can't

An API key is a scoped analysis credential, not a login. It shares your account quota, and nothing else about your account.

- ✓Run analyses.No-cost Starter scans (Reality Checks) and full paid deep scans, drawn from your account quota.
- ✓Poll and enrich.Run status, per-module states, idempotent enrichment of export sections.
- ✓Structured exports.Stable ideas-json (schemaVersion 2): scores, research, and the full deep report body on paid runs.
- ✓Deep modules.Proof of Demand, Founder Fit, Playbook, and Trends on your deep reports.
- ✓Share URLs.Optional public share links when you pass publish: true.

- ✗No dashboard access.A key cannot log into preuve.ai, read your account, or open the dashboard. It is not a session.
- ✗No account or billing control.Keys cannot change plans, buy tokens, manage payment methods, or create/revoke other keys.
- ✗API-created runs only.A key only sees analyses it created. Reports you ran in the web app stay invisible to it.
- ✗No profile writes.A founder profile submitted for Founder Fit is scoped to that report - it never rewrites your account preferences.

What is the Preuve startup validation API?

It is an HTTP API that runs a Preuve analysis on a startup idea and returns a viability score, market research, competitor analysis, a shareable report URL, and a structured JSON export. You authenticate each request with a single API key header, so any AI agent, CLI, or backend script can validate ideas programmatically.

How do I get a startup validation API key?

API keys are self-serve for paid Preuve accounts - generate yours instantly from the account panel (start on this page). Each key is scoped (analysis:write, analysis:read, batch:write, batch:read, export:read), sent in one x-preuve-key header, and tied to one Preuve account.

What is the difference between a starter and a deep scan?

scanType is explicit on every run. "starter" runs the Reality Check at no cost from your Starter scan allowance. "deep" runs the full paid analysis (15+ sections, pivots, citations) and consumes your account quota: subscription scans first, then lifetime monthly scans, then tokens. An agent can never spend paid quota by accident, and a refused deep scan fails loudly instead of silently downgrading.

Polling returns run status, enrichment progress, per-module states, and URLs. The export endpoint returns structured ideas-json (schemaVersion 2): scores, verdict, market, competitors, SWOT, risks, quick take, and citations on every tier, plus the full deep report body (sections, pivots, and generated module payloads) on paid deep runs owned by your key.

Optional add-on generations on paid deep reports, requested through the enrich endpoint: proofOfDemand (verified real-demand prospects), founderFit (founder-vs-plan underwriting from your founder profile), playbook (launch playbook), and trends (on-demand Google Trends). Each module is capped at one successful generation per report and is included in the deep report price - no extra fee.

Does an API key give access to my Preuve dashboard or account?

No. A key is not a login: it cannot open the dashboard, read your account, change billing, or manage other keys. It only sees analyses it created through the API - reports you ran in the web app stay invisible to it. If a key leaks, revoke it; your account itself is untouched.

API usage draws from the same analysis quota as your Preuve plan - Starter scans from your Starter allowance at no cost, deep scans from your subscription, lifetime, or token balance. There is no separate API fee, and failed runs refund whatever they claimed.

Should I validate before using an autonomous agent like Polsia?

Yes. Autonomous build agents such as Polsia execute fast but skip the demand-validation step, which is where most teams burn their runway. Run the idea through the Preuve validation API first, then hand a validated brief to your build agent. See the linked Polsia review for a deeper breakdown of that gap.