prompt-injection

What problem does it solve?

Protects against attacker-controlled text that hijacks LLM instructions. Direct injection arrives through user input; indirect injection arrives through retrieved documents, emails, or tool outputs. Both can cause the model to exfiltrate data, bypass guardrails, or execute unintended actions.

Core Features & Use Cases

• Detect vulnerable patterns where user data bleeds into system prompts
• Enforce boundaries by placing user data and retrieved documents in explicit segments
• Support RAG pipelines and external document processing with safe prompts

Quick Start

Identify and fix prompt construction patterns that allow user input to influence the system prompt, then reorganize prompts to separate developer instructions from data and wrap retrieved content with explicit delimiters.