liangzid/PromptExtractionEval

Source code of the paper "Why Are My Prompts Leaked? Unraveling Prompt Extraction Threats in Customized Large Language Models"

23
/ 100
Experimental

This project helps AI safety researchers and developers understand how easily confidential prompts can be extracted from customized Large Language Models (LLMs). It takes a fine-tuned LLM and a set of query inputs, then outputs an assessment of how vulnerable the model is to prompt leakage and the effectiveness of various defense strategies. This is for AI security engineers, red teamers, and researchers developing secure LLMs.

No commits in the last 6 months.

Use this if you need to evaluate the security of your customized LLMs against prompt extraction attacks and explore defense mechanisms.

Not ideal if you are looking for a general-purpose tool to prevent all types of data leakage or for immediate production-ready defense implementations.

LLM-security prompt-engineering AI-safety red-teaming model-vulnerability-assessment
No License Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 4 / 25
Maturity 8 / 25
Community 9 / 25

How are scores calculated?

Stars

7

Forks

1

Language

Python

License

Last pushed

Jul 23, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/prompt-engineering/liangzid/PromptExtractionEval"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.