DocShit – Sanitize your Documents Before they Reach LLMs

Summary: DocShit analyzes PDF and DOCX files to detect and remove prompt injections, hidden text, and malicious metadata, ensuring secure input for Large Language Models.

What it does

It scans documents to identify and neutralize embedded threats like prompt injections and harmful metadata before AI processing.

Who it's for

Users needing to secure documents against prompt injection and hidden content before feeding them to LLMs.

Why it matters

It prevents malicious content from compromising AI context and responses by sanitizing input documents.