Signature detection for contracts

Presence first

Contract operations teams lose time on files that were attached but never signed. A detection count answers that narrow question. It does not replace legal review, and it does not know your signature-block template. Pair it with a checklist of pages that must contain a mark.

The sample uses the PDF body. Field names and the page grouping are specified on PDF signature detection.

import base64
import requests

def detect_contract(path: str, api_key: str) -> dict:
    with open(path, "rb") as handle:
        pdf_base64 = base64.b64encode(handle.read()).decode("ascii")

    response = requests.post(
        "https://signature-detector.com/api/detect",
        headers={
            "Content-Type": "application/json",
            "x-api-key": api_key,
        },
        json={"pdfBase64": pdf_base64},
        timeout=120,
    )
    response.raise_for_status()
    data = response.json()

    return {
        "detection_id": data.get("detectionId"),
        "signed": data["signed"],
        "total_marks": data["count"],
        "credits_used": data["creditsUsed"],
        "pages": [
            {
                "page": page["page"],
                "count": page["count"],
                "boxes": [mark["bbox"] for mark in page["detections"]],
            }
            for page in data["pages"]
        ],
    }

Two boxes is not two parties

A signature page often has a company block and a counterparty block. You might sort boxes by y1 and assume the lower one is the customer. That is a layout guess. Write it down as a guess, show both crops to the clerk who files the contract, and correct it when the template changes. Cropping is local. The request does not return image bytes. See crop signatures from bounding boxes.

Workflow shape

• Receive the executed-copy PDF from email or the CLM.
• Detect, then list pages where count is 0.
• If a required page is empty, bounce the file. The longer version of this queue is automating signed-contract checks.
• If marks exist, store detectionId, page numbers, and boxes next to the contract record.
• For a no-code team, the same POST fits an HTTP step in n8n or Zapier.

Credits

• $10 per 100,000 credits, which is $0.0001 each. An image that is not served from cache spends 1 credit. A PDF spends 1 credit per page.
• 10 credits are added when the account is created.
• $1 minimum purchase. That is 10,000 credits. There is no subscription.
• Credits do not expire. The balance stays on the account until a detection spends it. A failed detection puts the credit back. Repeating an image or an identical PDF returns cached: true and does not spend a credit. On a PDF that hit is creditsUsed: 0.
• Buy credits from the dashboard or estimate a volume on the pricing page.

Sending the same PDF bytes again returns cached: true and creditsUsed: 0. The balance does not change. Re-encoding the file changes the hash, so that copy is charged at 1 credit per page.