Synthetic Document Generator turns a reference PDF into a shareable synthetic copy that preserves its layout and structure while replacing its text and personal information with realistic fictional content, then checks the result for remaining personal information. It is intended for Solutions Architects and proof-of-concept engineers creating documents for demonstrations, retrieval-augmented generation evaluations, model benchmarks, or partner handoffs.
aws-samples/sample-synthetic-document-generator
Agent Claude Code ✓ vendor
LLM grader for behavioral assertions on pocsynth skill evals. Reads a transcript and the eval's expected assertions, emits grading.json per the skill-creator convention.