The WAF here isn't very complicated, so it's no big deal for our intelligent agents to bypass it.
First, let's take a look at the training output (the image is randomly selected, and the generated payload may not necessarily trigger XSS).
Here, we've written a simple test page.
@chuanfua1
First, let's take a look at the training output (the image is randomly selected, and the generated payload may not necessarily trigger XSS).
Here, we've written a simple test page.
@chuanfua1
Static Data Flow Tracing (DF): Tracks the data flow of variables in a program to determine whether there is a data dependency between two variables or expressions. For example, in the "Risky First Deposit" vulnerability of Web 3, data flow analysis is used to determine whether the share is directly calculated using the deposit amount. (Similar to taint analysis of data flow, it checks whether there are variables controlled directly by users in the variables modified by the amount that have not been verified.)
@chuanfua1
@chuanfua1
If the modified payload is not intercepted by the WAF (e.g., the regular expression does not match the alert), the agent receives a 10-point reward.
The complexity of the training data (such as mixed-case characters and encoding methods) prompts the agent to explore more covert anti-detection methods and optimize the Q value.
The main training process involves the WAF environment and the agent engaging in a confrontation. The agent continuously learns to generate more effective payloads based on the WAF environment's feedback. After the WAF environment accepts these payloads, it conducts verification and returns a new state to the agent.
Let's take a look at the aforementioned WAF.
@chuanfua1
The complexity of the training data (such as mixed-case characters and encoding methods) prompts the agent to explore more covert anti-detection methods and optimize the Q value.
The main training process involves the WAF environment and the agent engaging in a confrontation. The agent continuously learns to generate more effective payloads based on the WAF environment's feedback. After the WAF environment accepts these payloads, it conducts verification and returns a new state to the agent.
Let's take a look at the aforementioned WAF.
@chuanfua1
We've already talked a lot about replaying the experience earlier, so there's not much to add here.
During the entire process, we trained for 100 rounds, and in each round, we first initialized and added 500 memories.
During training, we randomly selected a batch size of batch_size from this experience data for training.
@chuanfua1
During the entire process, we trained for 100 rounds, and in each round, we first initialized and added 500 memories.
During training, we randomly selected a batch size of batch_size from this experience data for training.
@chuanfua1
On XP, enter the location of the DLL file and press Enter repeatedly.
It shows "Doublepulsar Succeeded".
Check on Kali and find that the shell has already been reflected.
@chuanfua1
It shows "Doublepulsar Succeeded".
Check on Kali and find that the shell has already been reflected.
@chuanfua1
Global high-end quick kill data😂Support customized countries
With the most professional and efficient penetration technology overseas, and 4 years of penetration experience involving various global sites, our capabilities speak for themselves. Welcome to cooperate with us, Mr. Pan!
@chuanfua1
With the most professional and efficient penetration technology overseas, and 4 years of penetration experience involving various global sites, our capabilities speak for themselves. Welcome to cooperate with us, Mr. Pan!
@chuanfua1
National credit card data - real holdings and consumption
Real-time update of car owner data
Debit card data
Filter by: province/city/national level
Bank: can be specified
✅ Guaranteed authenticity, real holdings, and consumption
✅ Database kept up-to-date~ dynamic and active
✅ POS available, card application available, card usage available
@chuanfua1
Real-time update of car owner data
Debit card data
Filter by: province/city/national level
Bank: can be specified
✅ Guaranteed authenticity, real holdings, and consumption
✅ Database kept up-to-date~ dynamic and active
✅ POS available, card application available, card usage available
@chuanfua1
🔥🔥🔥Tag extraction from the app
Every 2 months is a deduplication cycle!
Specified extraction for the entire industry!
@chuanfua1
Every 2 months is a deduplication cycle!
Specified extraction for the entire industry!
@chuanfua1
每类漏洞都有针对性的审计技巧，在我们掌握了这些技巧之后，就可以有针对性地挖掘我们想要的漏洞。漏洞大致分为SQL注入、XSS、文件操作、代码/命令执行、变量覆盖以及逻辑处理，等等，这些都是常见的Web漏洞。
从事漏洞挖掘工作需要具备的知识是极其广泛的，并且随着时间在不断改变，也取决于您所研究的对象（Web程序、桌面程序、嵌入式等等）。不过，万变不离其宗，所需要掌握的知识领域却总可以认为是确定的，大致可以分为程序正向开发技术、攻防一体的理念、有效使用工具、对目标应用的理解四个方面。
@chuanfua1
从事漏洞挖掘工作需要具备的知识是极其广泛的，并且随着时间在不断改变，也取决于您所研究的对象（Web程序、桌面程序、嵌入式等等）。不过，万变不离其宗，所需要掌握的知识领域却总可以认为是确定的，大致可以分为程序正向开发技术、攻防一体的理念、有效使用工具、对目标应用的理解四个方面。
@chuanfua1
👍1