PromptShield

PromptShield is a local detection pipeline that analyzes incoming prompts for potential threats before they reach an LLM. It helps mitigate risks like prompt injection, jailbreaking, and off-topic requests by enforcing context boundaries.

This page includes a live interactive demo that connects directly to the PromptShield detection API on Railway.

Interactive Demo

Test your prompts against PromptShield's local detection pipeline.

1. Select Chatbot Context
2. Enter Prompt
Try an example:
3. Scan Result
Submit a prompt to see the scan contract.
Analyzing...
Threat Type
Confidence
Reason
Pipeline Layer
Scanner Models
Chatbot Model

🤖 Chatbot Response

Generating response...
$