Quickstart
Three verification engines behind one simple API: document forensics, AI-text detection, and citation fact-checking. No API key, no signup — your first call works from a terminal in under a minute.
Fact-check a report by URL
Point POST /v1/verify-references at any public report or article. Stipple fetches it, resolves every citation, recomputes the document's own arithmetic, and returns a structured verdict with a shareable permalink.
curl -X POST https://www.stipple.sh/v1/verify-references \
-H "Content-Type: application/json" \
-d '{"url": "https://example.com/report.pdf"}'The response is a trust summary plus per-item tables:
{
"summary": {
"references_total": 63,
"references_live": 57,
"references_archived": 3,
"references_dead": 3,
"papers_mismatched": 1,
"internal_issues": 0,
"headline": "60/63 citations resolve; 1 mis-attributed paper; 3 dead links"
},
"references": [ ... ],
"claims": [ ... ],
"internal_consistency": [ ... ],
"limitations": "Reports verification coverage, not truth. ...",
"check_id": "fc_6d22fb4033cbac40",
"permalink": "https://www.stipple.sh/fc/targeted-compliance-framework-review-6d22fb4033cbac40"
}Verify a document for tampering
Upload a PDF or image to POST /v1/warrants for forensic inspection — tampering signals, structure and metadata analysis, arithmetic reconciliation. You get back a warrant: a risk band, findings, and a shareable record.
curl -X POST https://www.stipple.sh/v1/warrants \
-F "file=@payslip.pdf"Detect AI-written text
Send prose (or a URL or file) to POST /v1/detect-ai-text. It returns a probability with the specific tells, and abstains on non-prose documents where style detection is unreliable.
curl -X POST https://www.stipple.sh/v1/detect-ai-text \
-H "Content-Type: application/json" \
-d '{"text": "Paste the prose you want to check here..."}'Limits while free
Anonymous use is rate-limited per IP: 20 documents/day, short bursts up to 10 requests, 25 MB max upload. A free API key (instant, no account) bumps you to 50/day on your own quota with usage metering. Details in Errors & limits.
Next steps
- API reference — every endpoint, parameter, and response shape.
- MCP integration — give Claude, Cursor, or your own agent these tools natively.
- agents.md — the machine-readable guide AI agents read directly.