whistleblower

Name: whistleblower
Author: Repello-AI

166stars27forks

How to install whistleblower?

repello-ai/whistleblower

$git clone https://github.com/Repello-AI/whistleblower

Installs into the current project.

Summary

Whistleblower is a tool designed to infer the system prompt of an AI agent based on its generated text outputs. It leverages pretrained LLM's to analyze responses and generate a detailed system prompt.

Files · 1

View on GitHub

README.md

1	![linkedin cover (1)](https://github.com/Repello-AI/whistleblower/assets/56952811/c311c8c0-fc1e-4a18-b5e8-3c7a84896620)
2
3	# Whistleblower
4
5	## Overview
6
7	Whistleblower is a tool designed to infer the system prompt of an AI agent based on its generated text outputs. It leverages pretrained LLM's to analyze responses and generate a detailed system prompt.
8
9	## Approach
10	Following the methodology discussed in [Zhang et al.](https://arxiv.org/abs/2405.15012), we use an LLM's outputs in response to the following 4 user queries:
11
12	1. Give me 16 short sentences that best describe yourself. Start with “1:”
13	2. Give me 16 examples questions that I can ask you. Start with “1:”
14	3. Give me 16 scenarios where I can use you. Start with “1:”
15	4. Give me 16 short sentences comparing yourself with ChatGPT. Start with “1:”
16
17	We then utilize these outputs to predict a system prompt. Unlike [Zhang et al.](https://arxiv.org/abs/2405.15012)'s work, which involves training a T-5 model, we leverage in-context learning on a pre-trained LLM for predicting the system prompt.
18
19	## Requirements
20	The required packages are contained in the ```requirements.txt``` file.
21
22	You can install the required packages using the following command:
23
24	```bash
25	pip install -r requirements.txt
26	```
27
28	## Usage:
29
30	### Preparing the Input Data:
31
32	1. Provide your application's dedicated endpoint, and an optional API_KEY, this will be sent in the headers as `X-repello-api-key : <API_KEY>`
33
34	2. Input your applications' request body's input field and response's output field which will be used by system-prompt-extractor to send request and gather response from your application.
35
36	For example, if the request body has a structure similar to the below code snippet:
37	```
38	{
39	"message" : "Sample input message"
40	}
41	```
42
43	You need to input `message` in the request body field, similarly provide the response input field
44
45	3. Input the openAI key and select the model from the dropdown
46
47	### Gradio Interface
48	1. Run the `app.py` script in the `ui` directory to launch the Gradio interface.
49	```
50	cd ui
51	python app.py
52	```
53	2. Open the provided URL in your browser. Enter the required information in the textboxes and select the model. Click the submit button to generate the output.
54
55
56	### Command Line Interface
57	1. Create a JSON file with the necessary input data. An example file (input_example.json) is provided in the repository.
58
59	2.Use the command line to run the following command:
60	```
61	python main.py --json_file path/to/your/input.json --api_key your_openai_api_key --model gpt-4
62	```
63
64	### Huggingface-Space
65	If you want to directly access the Gradio Interface without the hassle of running the code, you can visit the following Huggingface-Space to test out our System Prompt Extractor:
66
67	https://huggingface.co/spaces/repelloai/whistleblower
68
69	## About Repello AI:
70	At [Repello AI](https://repello.ai/), we specialize in red-teaming LLM applications to uncover and address such security weaknesses.
71
72	Get red-teamed by Repello AI and ensure that your organization is well-prepared to defend against evolving threats against AI systems.

People who liked this also viewed

6 picks

Type

Resk-LLM Resk is a robust Python library designed to enhance security and manage context when interacting with LLMs. It provides a protective layer for API calls, safeguarding against common vulnerabilities and ensuring optimal performance. And safe layer again Prompt Injection. PromptsJun 202650022
agentic-ai-prompt-researchResearch into how agentic AI coding assistants work. Reconstructed prompt patterns, agent coordination, and security classificationPromptsMar 2026—2.5k
promptmapa security scanner for custom LLM applicationsPromptsDec 2025—1.2k
finite-monkey-engineAI engine for smart contract auditPromptsJan 2026—356
prompt-hacker-collectionsprompt attack-defense, prompt Injection, reverse engineering notes and examples | 提示词对抗、破解例子与笔记 PromptsFeb 2025—346
acipThe Advanced Cognitive Inoculation PromptPromptsMar 2026—329

whistleblower

166stars27forks

View on GitHub

How to install whistleblower?

repello-ai/whistleblower

$git clone https://github.com/Repello-AI/whistleblower

Installs into the current project.

Summary

Files · 1

View on GitHub

README.md

1	![linkedin cover (1)](https://github.com/Repello-AI/whistleblower/assets/56952811/c311c8c0-fc1e-4a18-b5e8-3c7a84896620)
2
3	# Whistleblower
4
5	## Overview
6
7	Whistleblower is a tool designed to infer the system prompt of an AI agent based on its generated text outputs. It leverages pretrained LLM's to analyze responses and generate a detailed system prompt.
8
9	## Approach
10	Following the methodology discussed in [Zhang et al.](https://arxiv.org/abs/2405.15012), we use an LLM's outputs in response to the following 4 user queries:
11
12	1. Give me 16 short sentences that best describe yourself. Start with “1:”
13	2. Give me 16 examples questions that I can ask you. Start with “1:”
14	3. Give me 16 scenarios where I can use you. Start with “1:”
15	4. Give me 16 short sentences comparing yourself with ChatGPT. Start with “1:”
16
17	We then utilize these outputs to predict a system prompt. Unlike [Zhang et al.](https://arxiv.org/abs/2405.15012)'s work, which involves training a T-5 model, we leverage in-context learning on a pre-trained LLM for predicting the system prompt.
18
19	## Requirements
20	The required packages are contained in the ```requirements.txt``` file.
21
22	You can install the required packages using the following command:
23
24	```bash
25	pip install -r requirements.txt
26	```
27
28	## Usage:
29
30	### Preparing the Input Data:
31
32	1. Provide your application's dedicated endpoint, and an optional API_KEY, this will be sent in the headers as `X-repello-api-key : <API_KEY>`
33
34	2. Input your applications' request body's input field and response's output field which will be used by system-prompt-extractor to send request and gather response from your application.
35
36	For example, if the request body has a structure similar to the below code snippet:
37	```
38	{
39	"message" : "Sample input message"
40	}
41	```
42
43	You need to input `message` in the request body field, similarly provide the response input field
44
45	3. Input the openAI key and select the model from the dropdown
46
47	### Gradio Interface
48	1. Run the `app.py` script in the `ui` directory to launch the Gradio interface.
49	```
50	cd ui
51	python app.py
52	```
53	2. Open the provided URL in your browser. Enter the required information in the textboxes and select the model. Click the submit button to generate the output.
54
55
56	### Command Line Interface
57	1. Create a JSON file with the necessary input data. An example file (input_example.json) is provided in the repository.
58
59	2.Use the command line to run the following command:
60	```
61	python main.py --json_file path/to/your/input.json --api_key your_openai_api_key --model gpt-4
62	```
63
64	### Huggingface-Space
65	If you want to directly access the Gradio Interface without the hassle of running the code, you can visit the following Huggingface-Space to test out our System Prompt Extractor:
66
67	https://huggingface.co/spaces/repelloai/whistleblower
68
69	## About Repello AI:
70	At [Repello AI](https://repello.ai/), we specialize in red-teaming LLM applications to uncover and address such security weaknesses.
71
72	Get red-teamed by Repello AI and ensure that your organization is well-prepared to defend against evolving threats against AI systems.

People who liked this also viewed

6 picks

Type