Lenticular is an advanced PDF analysis and manipulation framework designed for security researchers, penetration testers, and red teams. It provides a local, web-based interface to inspect internal PDF structures, extract raw data, and inject custom payloads for vulnerability testing, with a specific focus on evaluating Large Language Models (LLMs) and automated parsing systems against prompt injection attacks.
- Raw Inspection: Navigate and analyze the internal PDF object hierarchy, metadata (XMP), and hidden structures.
- Text Extraction: Decompress and extract readable text directly from PDF data streams.
- Stealth Injection: Embed OCR-based payloads using imperceptible text rendering techniques to bypass visual inspection while remaining readable to machine parsers.
- Dictionary Injection: Inject custom logic and payloads directly into PDF dictionary objects (e.g.,
/LenticularLogic) to test parser robustness.
The framework is built on a modular architecture to ensure stability and cross-platform compatibility:
- Backend: Python-based microservice utilizing Flask, PyMuPDF (fitz), and pikepdf for high-performance document manipulation.
- Frontend: A responsive web GUI featuring Split View rendering and Monaco Editor integration for real-time, IDE-grade code analysis.
Ensure you have Python 3.x installed along with the required dependencies.
pip install flask pikepdf pymupdfAccess the interface via your standard web browser at http://127.0.0.1:5005.
Legal Disclaimer
Lenticular is developed strictly for educational purposes, authorized security research, and defensive engineering. Users are solely responsible for their actions. The authors and contributors are not responsible for any misuse, damage, or legal implications caused by the use of this software. Always obtain explicit permission before testing external systems.