Associating concern for that whole defrauding of document changing across platform its particular center, getting your break-in and misuse of vital doc. To help mitigate this problem, a new document authentication system, called ForgeryGuard, based on machine learning that detects forgery by performing an in-depth analysis of document structure and content is introduced. ForgeryGuard implements a TensorFlow-based MLP model of real-time signature forgery detection based on smart feature extraction and data storage processes. The proposed system extracts nine effective sequential features from the signature images, which include ratio, centroids, eccentricity, solidity, skewness, and kurtosis, that are both spatial and geometric. Such features are stored in the ForgeryGuard test feature repository in CSV format in a systematic manner for a simpler testing process. The MLP architecture with three hidden layers provides classification for the features that were extracted and classifies them as either real or forge class signatures. The model, trained using the Adam optimizer for 1000 epochs, got a testing accuracy of 91% and EER 4.5%. The system also features signature image processing in an average of 0.35 s per document, enabling high accuracy and near real-time detection performance. Powered by its scalable and well-tested architecture, ForgeryGuard shines as a breakthrough technology for fighting against document forgery in sensitive and high-impact environments to help keep the digital world secure.