Truedocs - API para validacion de documentos

Extract Document Text

Extract text from documents (images/PDFs). This tool uses AWS Textract to extract text and metadata from document images and PDFs. Supports various image formats (JPEG, PNG, TIFF, etc.) and PDF files. When `fields` is provided (a non-empty list of field names), extraction is routed to a vision model instead of Textract, which returns only the requested fields. Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR doc

Inspect Document

Inspect the image FORMAT of a document (measurements, not verdicts). Returns deterministic, per-pixel measurements read from the ORIGINAL bytes: container type (image/pdf/docx), page/embedded-image counts, orientation (portrait/landscape), and per sub-image detail — dimensions, aspect ratio, DPI, colour mode, mean saturation, EXIF orientation, and (for PDF embedded images) the on-page placement size in centimetres. Use this to check image-format rules such as: is the document vertical, is it i

Classify Document

Classify document type automatically. Uses machine learning models to identify the type of document from Mexican government and financial institutions. Supports various document types including INE, RFC, utility bills, and more. Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR document) confidence_threshold: Minimum confidence level for classification (0.0-1.0) Returns: Document classification with confi

Verify Document Sync

Verify document authenticity synchronously. Performs real-time verification of Mexican government documents against official databases. Supports INE, RFC, and other official documents. Returns verification results immediately. Args: document_type: Type of document (e.g., "ine", "constancia_de_situacion_fiscal") document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR document) rfc: RFC number for additional verificatio

Verify Document Async

Verify document authenticity asynchronously. Starts an asynchronous verification process for complex document validation that may take longer. Returns a job ID that can be used to check progress and retrieve results when complete. Args: document_type: Type of document (e.g., "ine", "constancia_de_situacion_fiscal") document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR document) callback_url: URL to notify when verif

Check Job Status

Check the status of an asynchronous verification job. Retrieves the current status and results (if complete) of an async verification job. Jobs progress through states: pending -> in_progress -> completed/failed. Args: job_id: ID of the job to check (returned from async verification) Returns: Job status, progress, and results if available Example: check_job_status("550e8400-e29b-41d4-a716-446655440000")

Detect Forgery Metadata

Analyze document metadata and structure for forgery indicators (Phase 1). Performs fast, deterministic forensic analysis of PDF metadata and structure without invoking any LLM. Checks for suspicious producers (Canva, Photoshop, Word), incremental updates, multiple EOF markers, flat image exports, and embedded JavaScript. Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR document) Returns: riskScore (0-100), r

Detect Forgery Visual

Detect visual anomalies in a document using AI vision (Phase 2). Uses a vision language model to spot visual forgery artifacts such as text overlay artifacts (kerning, baseline misalignment), inpainting regions (unnatural smoothing, blurred patches), compression mismatches, copy-move forgery, and rendering inconsistencies. Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR document) Returns: findings[] with re

Detect Forgery Semantic

Check a document for semantic and logical inconsistencies (Phase 3). Uses an LLM to verify internal consistency: math (line items sum to subtotal), dates (issue date not in the future, period consistency), address/identity (header matches footer), numerical (unusual precision, balances reconcile), and reference consistency (invoice sequences, page numbers). Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to document (provide this OR documen

Detect Forgery

Perform a full composite forgery assessment on a document. Runs all three detection phases (metadata, visual, semantic) in a single request and returns a synthesized verdict. Use this as the primary tool when you need an overall authenticity judgment. Scoring weights: metadata 40%, semantic 35%, visual 25%. Verdicts: likely_authentic (0-25), suspicious (26-60), likely_forged (61-100). Args: document: Base64 encoded document data (provide this OR document_url) document_url: URL to docu

Health Check

Check the health and connectivity of the Truedocs API. Performs a simple health check to verify that the API is accessible and authentication is working correctly. Useful for debugging connectivity issues. Returns: Health status and API information Example: health_check()