Security Test System

The security test system uses LLMs to simulate human-like interactions and manage sequential operations, addressing inefficiencies in web application security testing by optimizing patrol procedures and reducing redundant visits, thereby enhancing vulnerability detection efficiency.

JP7580005B1Active Publication Date: 2024-11-08UB SECURE CO LTD

Patent Information

Application Number
JP2024096407
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-06-14
Publication Date
2024-11-08
Estimated Expiration
2044-06-14

AI Technical Summary

Technical Problem

Existing web application security testing systems struggle to efficiently simulate sequential operations on web pages, leading to inefficiencies in identifying vulnerabilities due to the complexity and scale of possible operation procedures.

Method used

A security test system that utilizes Large-Scale Language Models (LLM) to create operation procedure manuals, simulating human-like interactions and efficiently handling sequential operations across web pages, while reducing redundant visits by identifying changes in web page behavior.

Benefits of technology

The system effectively grasps appropriate operating procedures for websites with sequential operations, enhancing the efficiency and effectiveness of automatic patrols by minimizing redundant visits and improving the comprehensiveness of vulnerability detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007580005000001_ABST
    Figure 0007580005000001_ABST
Patent Text Reader

Abstract

It automatically crawls websites that include sequential operations efficiently and effectively by understanding the appropriate operating procedures. [Solution] A security test system 1 that checks for security vulnerabilities in web applications, retrieves and analyzes a web page to be traversed within a target website 3, sets information related to the content of the retrieved web page as a prompt and inputs it to an LLM 4, creates operation procedure information related to the web page, simulates operations related to the web page in accordance with the operation procedure information, retrieves URLs related to links in the web page and registers them in a list of pages to be traversed 14.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an application testing technique, and more particularly to a technique that is effective when applied to a security test system that checks for vulnerabilities in Web applications. [Background technology]

[0002] Web applications are premised on using networks, and from a security perspective, it is extremely important to inspect and test for vulnerabilities. There are a variety of tools and services available for inspecting web applications for vulnerabilities, and these are being investigated and developed on a daily basis.

[0003] Web application security testing methods can be broadly divided into SAST (Static Application Security Testing) and DAST (Dynamic Application Security Testing). While SAST statically analyzes source code, etc., DAST sends simulated attack (inspection) requests to a running application from an attacker's perspective and determines whether there are vulnerabilities from changes in the application's behavior. Therefore, in a DAST security testing system, it is necessary to identify the pages in the web application that will be the target of the attack (inspection), and for this purpose, the target web application (website) is crawled automatically or manually, and the configuration of links, etc. is analyzed to collect information about the pages.

[0004] As an example of a technology related to the automatic crawling of such websites, Patent Gazette No. 7320211 (Patent Document 1) describes a method for automatically crawling a website in a vulnerability inspection of the website, using AI (Artificial Intelligence) to determine functions that require inspection for vulnerabilities, and determining the degree of association between multiple executable operations that can be performed on a web page and the functions that require inspection, and executing operations identified as having a high degree of association with them as a priority. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 7320211 Summary of the Invention [Problem to be solved by the invention]

[0006] According to the above-mentioned conventional technology, when automatically crawling a website, among multiple operations that can be performed on each web page, the operations can be executed starting with the highest priority operation. Therefore, it is said that it is possible to prevent important functions from being overlooked even if an upper limit is set based on the number of web pages to be automatically crawled, the number of hierarchical levels, the elapsed time, etc.

[0007] However, there are cases where a website includes operations that require a sequence, such as the next operation being impossible until a specific operation has been performed. For example, on an EC (Electronic Commerce) website, there are cases where the order screen cannot be accessed until the customer's address is entered, or the order screen cannot be accessed until an item has been added to a cart. In such cases, communication will not occur unless the operations are performed in the correct order. Therefore, in order to fully understand all the operations that can be performed on each web page, it is necessary to correctly understand the functions of the website and simulate operations in the correct order.

[0008] In this regard, it is difficult to determine in advance all the patterns of what order operations should be performed in what cases, for example, as rules. On the other hand, it is theoretically possible to determine the appropriate operation procedure by trying all possible operations in a brute force manner. However, depending on the scale and complexity of the web page, the number of combinations of operation procedures that must be tried can become enormous, making it difficult to simulate the operations in a realistic amount of time.

[0009] Therefore, an object of the present invention is to provide a security test system that can grasp appropriate operation procedures for a Web site that includes operations that require sequence, and can automatically crawl the Web site efficiently and effectively.

[0010] The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings. [Means for solving the problem]

[0011] Among the inventions disclosed in this application, a brief summary of representative inventions is as follows.

[0012] A security test system that is a representative embodiment of the present invention is a security test system that checks for security vulnerabilities in a Web application, and acquires and analyzes a Web page to be traversed within the Web application to be tested, sets information related to the content of the acquired Web page as a prompt and inputs it into an LLM (large scale language model) to create first operation procedure information related to the Web page, simulates operations related to the Web page in accordance with the first operation procedure information, acquires URLs related to links in the Web page, and registers them in a list of pages to be traversed. Effect of the Invention

[0013] The effects obtained by the representative inventions disclosed in this application can be briefly explained as follows.

[0014] In other words, according to a representative embodiment of the present invention, in a mechanism for inspecting vulnerabilities in a website, it is possible to understand appropriate operating procedures for a website that includes operations that require sequence, and to automatically patrol the website efficiently and effectively. [Brief description of the drawings]

[0015] [Figure 1] 1 is a diagram showing an overview of a configuration example of a security test system according to an embodiment of the present invention. [Diagram 2] 1 is a flowchart outlining an example of the flow of an automatic patrol process in one embodiment of the present invention. [Diagram 3] FIG. 2 is a diagram showing an overview of a specific example of a patrol target page list in one embodiment of the present invention. [Figure 4] FIG. 1 is a diagram showing an overview of an example of a Web page according to an embodiment of the present invention. [Diagram 5] FIG. 11 is a diagram showing an overview of another example of a Web page according to an embodiment of the present invention. [Figure 6] FIG. 1 is a diagram outlining an example of creating an operating procedure manual using LLM in one embodiment of the present invention. [Figure 7] FIG. 13 is a diagram showing an overview of another specific example of the crawl target page list in one embodiment of the present invention. [Figure 8] FIG. 13 is a diagram outlining an example of determining which Web pages need to be re-routed using LLM in one embodiment of the present invention. [Figure 9] FIG. 13 is a diagram outlining an example of determining which Web pages need to be re-routed using LLM in one embodiment of the present invention. [Figure 10] 1 is a diagram showing an overview of an example of the data configuration of a patrol target page list in one embodiment of the present invention. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. In all the drawings for explaining the embodiment, the same parts are generally given the same reference numerals, and the repeated explanations will be omitted. On the other hand, a part that has been described with a reference numeral in a certain drawing may be referred to with the same reference numeral in the explanation of another drawing, although it will not be shown again.

[0017] <Summary> When testing a website security using DAST, it is necessary to crawl the target website to find all publicly available URLs (Uniform Resource Locators) (i.e., the URLs to be inspected). (More precisely, for URLs that have common input / output processing rather than all communications that occur, the results will not change if each URL is inspected individually, so it is sufficient to inspect (crawl) at least one URL, which means all URLs that have different input / output processing.)

[0018] To find the URL, there are static methods that analyze the HTML (HyperText Markup Language) of the website, and dynamic methods that simulate actual operations on the website and collect the traffic that occurs. When using the latter method, as mentioned above, there are cases where the website contains sequential operations, and the next operation cannot be performed until a specific operation has been performed, and traffic will not occur unless the operations are performed in the correct sequence. Therefore, to simulate actual operations, it is necessary to correctly understand the functions of the website and simulate the operations in the correct sequence.

[0019] In this regard, it is difficult to determine in advance all the patterns of what order operations should be performed in what cases, for example, as rules. On the other hand, it is theoretically possible to determine the appropriate operation procedure by trying all possible operations in a brute force manner. However, depending on the scale and complexity of the web page, the number of combinations of operation procedures that must be tried can become enormous, making it difficult to simulate the operations in a realistic amount of time.

[0020] Here, websites that include sequential operations as described above may include two types of web pages: those that require multiple operations to be performed in an appropriate order within a web page, and those that require operations to be performed in an appropriate order across multiple web pages.

[0021] An example of the type of situation where multiple operations need to be performed in the correct order on a Web page is when, on an e-commerce site, on a screen for registering a delivery address for a product, if you press the "Register" button without filling in the name and address fields, an error occurs prompting you to enter your name and address, and you are unable to proceed to the registration completion screen (to proceed to the registration completion screen, you must enter your name and address before pressing the "Register" button).

[0022] Another example of the type of situation where operations need to be performed in the appropriate order across multiple Web pages is when, for example, on an e-commerce site, if you go to the cart screen before adding an item to the cart, the cart is empty so the "Proceed to checkout" button is not displayed and you cannot proceed to the order screen; however, if you add an item to the cart and then go to the cart screen, there is an item in the cart so you can press the "Proceed to checkout" button and proceed to the order screen.

[0023] In a security testing system which is one embodiment of the present invention, in order to accommodate the former type of web page where multiple operations need to be performed in an appropriate order within the web page, in addition to or instead of a method (conventional method) of obtaining appropriate operating procedures based on predefined rules for what order to perform operations in what cases, appropriate operating procedures (operating procedure manual) are created using a method that utilizes generative AI / Large Language Models (hereinafter sometimes collectively referred to as "LLM") such as ChatGPT (registered trademark) to obtain operating procedures that are close to the content that would be obtained if operated by a human.

[0024] In addition, to deal with the latter type of web page that requires operations to be performed in an appropriate order across multiple web pages, the LLM determines whether or not it is predicted that the behavior of the target web page will change between when it was visited in the past and when it is visited again in the present, and if it is predicted that a change will occur, it is considered that the operation procedures are different even if it is the same URL and is individually subject to visiting (inspection). Note that, as will be described later, in this embodiment, by grouping and handling web pages that are essentially the same but have different URLs, the number of web pages to be visited is substantially reduced and the efficiency of the process of visiting and inspection is improved, but if a web page is determined to be individually subject to visiting because its behavior has changed from when it was visited in the past even if it is the same URL, it is removed from the grouping and subject to re-visiting.

[0025] <System configuration> 1 is a diagram showing an overview of an example of the configuration of a security test system according to an embodiment of the present invention. The security test system 1 is configured, for example, by a server device or a virtual server constructed on a cloud computing service, and a user terminal 2 such as a PC (Personal Computer) used by a user accesses the system using a web browser or a dedicated application (not shown) via a network such as the Internet, a VPN (Virtual Private Network), or a LAN (Local Area Network) (not shown).

[0026] The security test system 1 realizes various functions related to the implementation of security tests by, for example, executing an operating system (OS), a database management system (DBMS), middleware such as a web server program, and software that runs on the OS and the DBMS, which are expanded onto memory from a recording device such as a hard disk drive (HDD) or solid state drive (SSD), using a central processing unit (CPU) (not shown). The security test system 1 has various units, such as an inspection management unit 11, a patrol processing unit 12, and an inspection execution unit 13, which are implemented by software. It also has various data stores, such as a patrol target page list 14, an operating procedure manual creation rule 15, a page list 16, and an inspection result 17, which are implemented by a database, a file table, or the like.

[0027] The inspection management unit 11 has management functions and user interface functions related to vulnerability inspection, such as accepting input and settings of information related to the target website 3 to be inspected from the user via the user terminal 2, accepting input of various settings and instructions related to the implementation of the inspection, and creating a report based on the inspection result 17, which is the result of the implementation of the inspection, and presenting it to the user via the user terminal 2.

[0028] The crawling processor 12 has a function of automatically crawling a target website 3 specified by a user, for example, registering the detected web pages in a crawl target page list 14, and creating a page list 16 including the web pages to be inspected. As described above, the crawling processor 12 grasps appropriate operation procedures for web pages that include sequential operations included in the target website 3 by using the operation procedure manual creation rules 15 and the LLM 4, and creates an operation procedure manual for the target web pages. The details of the processing contents of the crawling processor 12 will be described later.

[0029] The test execution unit 13 has a function of, for example, testing each Web page registered in the page list 16 for vulnerability using the DAST method. That is, for each target Web page, when sending a request to the target website 3, it sends a pseudo attack request by, for example, inputting an invalid value into a form. Then, it analyzes whether the response from the target website 3 is an unusual response to determine whether there is a vulnerability, and records the determination result in the test result 17.

[0030] <Processing flow> 2 is a flow chart outlining an example of the flow of the automatic patrol process by the patrol processing unit 12 in one embodiment of the present invention. When the automatic patrol process is started, the patrol processing unit 12 first obtains the start page of the target Web site 3 designated by the user and registers it in the patrol target page list 14 (S01). Then, it is determined whether or not the end condition of the automatic patrol is met (S02). For example, if the patrol target page list 14 is empty, or the number of patrolled pages reaches a preset upper limit, or the elapsed time from the start of patrol reaches a preset timeout time, it is determined that the end condition is met (Yes in step S02), and the automatic patrol process is ended.

[0031] If the termination condition is not met (No in step S02), a web page (URL) to be visited is determined from the unvisited pages based on the list of pages to be visited 14 (S03), and that page on the target website 3 is accessed and analyzed (S04).

[0032] FIG. 3 is a diagram outlining a specific example of the patrol target page list 14 in one embodiment of the present invention. Among the URLs in the list, those with a patrol status of "patrolled" are not patrolled because they have already been patrolled. Also, those with a patrol status of "URL similar to XX" are not patrolled because they are similar to the "patrolled" URL (as described above, in this embodiment, Web pages that are different URLs but substantially the same are grouped and handled to reduce the number of Web pages to be patrolled). In the example of FIG. 3, the next URL to be patrolled is, for example, URL "No. 6" because it was found first among those with a patrol status of "not patrolled".

[0033] Returning to FIG. 2, the target Web page is statically analyzed (S04), and then a rule-based operating procedure manual is created for appropriate operating procedures (S05). In this embodiment, an operating procedure manual refers to data configured to include information on one or more operation contents in the target Web page (URL) (for convenience, it is called an "operating procedure 'manual', but the actual entity is data (operating procedure information)). There may be multiple operating procedures manuals for one Web page, and when the operation contents of one operating procedure manual are completed, the target Web page is reloaded before moving on to the execution of the operation of the next operating procedure manual so that the operation contents do not interfere with each other during simulation.

[0034] An example of how to create a rule-based operating procedure manual is as follows: First, in step S04 of Fig. 2, the HTML of the target Web page is analyzed to ascertain executable operations based on event listeners and the like registered in elements. At this time, operations for elements that should be operated at once, such as elements enclosed in HTML Form tags, are grouped.

[0035] Note that "elements that should be operated at once" here does not only refer to elements within a Form tag, but also includes elements that require multiple operations, such as elements that display a drop-down menu when clicked, or drag and drop. Therefore, not only multiple operations on one element, but also operations on multiple different elements may be grouped together. These rules are registered in Operation Procedure Creation Rules 15. After that, the specific operation content (selection values, input contents, etc.) for each element is determined and an operating procedure manual is created.

[0036] 4 is a diagram outlining an example of a Web page in an embodiment of the present invention. In this example, the elements "Postal Code," "Prefecture," "City / Ward / Town / Village," "Back," and "Register" are arranged in Form tags on the "Address Registration" screen below the "Top" screen. Operations that can be performed on this screen can be understood as follows: "Top" is an a tag and so can be clicked; "Postal Code" is an input tag and so can be entered; "Prefecture" is a select tag and so the selected value can be changed; "City / Ward / Town / Village" is an input tag and so can be entered; and the "Back" and "Register" buttons are each set with a click event and so can be clicked.

[0037] Among these elements, the operations related to "Top" and "Back" are not grouped, while the operations related to the elements within the Form tag, "Postal Code", "Prefecture", "City / Ward", and "Register", are grouped together. Note that although "Back" is an operation within the Form tag, it is an operation that causes a screen transition, so it will be separated in the operating procedure manual.

[0038] After that, the specific operation content for each element is determined and an operation procedure manual is created. In the example of FIG. 4, for example, "operation procedure manual 1" is created with the content of clicking the "Top" link. Also, "operation procedure manual 2" is created with the content of entering "123-4567" in "postal code", selecting "Tokyo" as "prefecture", entering "Chuo-ku" in "city, ward, town, village", and clicking the "Register" button. Also, "operation procedure manual 3" is created with the content of clicking the "Back" button. Note that the specific setting values, etc. can be determined by an appropriate method, for example, by registering a pattern in the operation procedure manual creation rules 15.

[0039] Returning to FIG. 2, in addition to creating a rule-based operating procedure manual (S05), in this embodiment, LLM4 is used to create an operating procedure manual that is close to the content when operated by a human (S06).

[0040] Fig. 5 is a diagram outlining another example of a Web page in an embodiment of the present invention. This shows an example of a screen for registering a delivery address on an EC site, and is designed so that the delivery address cannot be registered unless the required fields (marked with *) for name, address, and phone number are filled in beforehand. Note that no Form tag is used here.

[0041] Here, in creating an operating procedure manual based on rules in step S05 of FIG. 2, multiple operations that correspond to the rules set in the operating procedure manual creation rules 15 as "elements that should be operated at once," such as elements within a Form tag, can be grouped into one operating procedure manual as described above.

[0042] On the other hand, some websites do not use the standard HTML Form tag, but implement it with their own tags, as in the example in Figure 5, and the "elements that should be operated at once" cannot be correctly recognized, and the operation instructions may be separated. For example, when creating operation instructions based on the rules in step S05 in Figure 2 for the example web page in Figure 5, "Operation Instructions 1" is created with the content of inputting "○○" into the "Last Name" of "Name", "Operation Instructions 2" is created with the content of inputting "△△" into the "First Name" of "Name", "Operation Instructions 3" is created with the content of inputting "●●" into the "Sei" of "Name (Kana)", and so on.... (omitted)... "Operation Instructions X" is created with the content of clicking the "Register" button, and so on.

[0043] As a result of the separated operating instructions, for example, even if the "Last Name" of "Name" is entered in "Operating Instruction 1," the target Web page is reloaded before "Operating Instruction 2" is executed, so the user is returned to a state where the "Last Name" has not been entered and the operation to enter the "First Name" begins, resulting in a problem in which the operations that should be performed in one go (in the example of Figure 5, entering information into a required field) cannot be performed correctly.

[0044] In contrast, in the creation of operating instructions using LLM4 in step S06, operating instructions that are closer to the content that a human would use in cases where a function equivalent to a Form tag is implemented with a unique tag, or where the creation of a rule-based operating instructions in step S05 requires complex operations for which appropriate operating instructions cannot be created for "elements that must be operated at once," are created. This makes it possible to create "operating instructions 1" with the content, for example, of entering "○○" in the "surname" field of the required "name," entering "△△" in the "first name," entering "●●" in the "name (kana)" field, and then clicking the "Register" button.

[0045] FIG. 6 is a diagram outlining an example of creating an operating procedure manual using LLM4 in one embodiment of the present invention (step S06 in FIG. 2). Here, by setting the HTML content of a screen such as the example in FIG. 5 as a prompt template, a prompt for causing LLM4 to create an operating procedure manual is obtained, as shown in the upper diagram in FIG. 6. Note that while the HTML information of the target page is entered in the example in FIG. 6, information obtained by screen capture can also be used. Also, it is not necessary to enter the entire HTML text; it may be entered after a trimming process to remove unnecessary information.

[0046] By inputting an operating procedure creation prompt such as that shown in the upper figure into LLM4, an operating procedure (operating procedure information in JSON (JavaScript Object Notation) format, etc.) such as that shown in the lower figure can be obtained. By simulating operations using the obtained operating procedure, it becomes more likely that requests that could not be found by simulating operations using an operating procedure obtained from a rule base (for example, a request to register a new address in the example screen of Figure 5) can be found.

[0047] In this embodiment, the comprehensiveness of operations in the operating procedure manual is efficiently improved by performing both the conventional rule-based creation of the operating procedure manual (step S05 in FIG. 2) and the creation of the operating procedure manual using LLM4 (step S06 in FIG. 2), but it is also possible to only create the operating procedure manual using LLM4. Also, in the example of FIG. 2, it is described that the steps are executed in the order of rule-based creation of the operating procedure manual → creation of the operating procedure manual using LLM4, but the steps may be executed in the reverse order or in parallel.

[0048] Returning to Fig. 2, the screen operations of the target Web page are then simulated according to the operation manual created in steps S05 and S06 (S07). Then, it is determined whether the screen display has changed as a result of the operations (S08). If the screen has changed (Yes in step S08), information about the changed screen (e.g., DOM (Document Object Model) or a screen capture) is used to return to step S05 to create an operation manual (rule-based and LLM4-based) and to repeat the simulation of screen operations according to the created operation manual.

[0049] On the other hand, if there is no longer any change in the screen display due to the operation (No in step S08), a list of pages to be visited 14 is created from a list of URLs found in the target Web page (S09). At this time, if the target Web page has already been visited, there is a high possibility that the URLs found will be the same even if the Web page is visited again. Therefore, in this embodiment, in order to reduce unnecessary visits, a rule base is used to determine whether the found URL overlaps (is substantially the same as) an existing Web page that has been visited in the past, and the overlapping ones are grouped and excluded from the visit (S10).

[0050] However, even if a URL is substantially the same as an existing Web page that has been visited in the past, if the preconditions (operations performed) to reach the URL are different, the behavior of the target Web page may change, and in that case, it is necessary to visit the target Web page again. Therefore, in this embodiment, LLM4 is used to determine whether or not the behavior of the target Web page is expected to change from when it was visited in the past, and if it is expected to change, the target Web page is removed from the grouping target and made the target Web page for re-visiting (S11).

[0051] 7 is a diagram outlining another specific example of the crawl target page list 14 in one embodiment of the present invention. In the example of FIG. 7, the "No." column indicates the order in which each URL was found, and the "Parent" column indicates the number of the parent screen in which the link of the target URL was found. That is, it shows that by analyzing the top screen "No.1" in step S09 of FIG. 2, a cart screen "No.2" and a product list screen "No.3" were found. Then, by analyzing the screen "No.3" a screen for adding a product "No.4" to the cart was found, and by analyzing the screen "No.4" a cart screen "No.5" was found.

[0052] At this time, in the process of step S10 in Fig. 2, since the cart screen of "No. 5" is similar (substantially identical) to the cart screen of "No. 2" and has already been visited, it is grouped with the screen of "No. 2" and it is determined that visiting is not necessary. However, since the cart screen of "No. 2" shows the cart in its initial state, with no products added to the cart and it is empty, it is not possible to proceed to the checkout screen, but since the cart screen of "No. 5" transitioned after a product was added to the cart on the screen of "No. 4," it is expected that a new operation (URL) to proceed to the checkout will be found by visiting again (i.e., these Web pages contain operations that are ordered across multiple Web pages).

[0053] Therefore, in this embodiment, by processing step S11 in Figure 2, taking into account the prerequisites (operations) for reaching the target URL from the list of URLs grouped based on the rule base in step S10, LLM4 is used to determine whether or not it is predicted that the behavior of the Web page will change between the previous visit (screen "No. 2") and the current visit (screen "No. 5"), i.e., whether or not a re-visit is necessary.

[0054] Figures 8 and 9 are diagrams outlining an example of determining which Web pages need to be re-routed using LLM4 (step S11 in Figure 2) in one embodiment of the present invention. In this example, by setting URLs that have already been found (routed) and URLs found in the current roaming in the contents of the list of pages to be roamed 14 such as the example in Figure 7 in the prompt template, a prompt is obtained for causing LLM4 to determine which URLs need to be re-routed, as shown in Figure 8.

[0055] Here, in the data of the visited URL, in addition to the screen ID ("id") and parent ID ("parentId") properties, a text showing an overview of the operations is included in the "summary" property, but it is also possible to include not only an overview of the operations performed in this visit, but all operations performed since the start of the visit. For example, in the example of Figure 7 above, after adding a product to the cart on screen "No.4", the user is moved directly to cart screen "No.5". However, if the website is structured such that after adding a product to the cart, the user cannot move to the cart screen without first returning to the top screen "No.1", it is not possible to properly determine whether or not the "No.5" cart screen needs to be visited again without tracing back the process of moving to the top screen.

[0056] By inputting a prompt as shown in the example of Fig. 8 into the LLM4, it is possible to obtain an output as shown in Fig. 9. This output includes information on URLs that require re-routing. Note that in this embodiment, whether or not re-routing is required is determined using the LLM4 as described above, but in cases where there are a large number of URLs to be determined, in order to improve efficiency, some or all of them may be determined using a rule base instead of the determination by the LLM4.

[0057] 2, the URL that was determined to need to be re-routed in step S11 is registered in the list of pages to be visited 14 (or, if the target URL is already registered in the list of pages to be visited 14, its visiting status is changed) (S12). Then, the process returns to step S02 and the subsequent processes are repeated until the end condition for automatic roaming is met (Yes in step S02).

[0058] <Data Structure> 10 is a diagram outlining an example of the data configuration of the crawl target page list 14 in one embodiment of the present invention. The crawl target page list 14 is a table that holds a list of pages (URLs) that are the subject of (or have been crawled) automatic crawling, and has items such as the crawl target page ID, parent page ID, crawl request ID, URL, URL summary, method, request, response, capture, page crawl status, operation content, and operation summary.

[0059] The "Tour target page ID" field holds ID information that uniquely identifies the page to be toured. The "Parent page ID" field holds ID information that identifies the parent page in which the tour target page was found. The "Tour request ID" field holds ID information that uniquely identifies the tour request for the target website 3 that contains the target page. By making it possible to individually identify each tour request, for example, information on the overall status of each tour (waiting to be toured, tour in progress, tour completed, etc.) can be managed separately in a management table (not shown) or the like.

[0060] The URL item holds information about the URL of the target page. The URL summary item holds text obtained as a result of analyzing the target page that gives an overview of what the target page is (for example, "Home page" or "Add product to cart") The Method, Request, Response, and Capture items respectively hold the method of the target page (GET, POST, PUT, etc.), the request for the target page and the response to this, and a screen capture of the target page. The Page Crawl Status item holds information about the crawling status of the target page (waiting to crawl, crawling, crawling completed, etc.).

[0061] The operation content item holds the specific operation content required to reach the URL of the target page (e.g., [{"name":"open","url":"https: / / example.com / top / "},{"name":"clickAt","target":" / / a[.='link']"}], etc.). The operation summary item holds a sentence showing the summary of the above operation content (e.g., "Open the top screen and click [link]"), etc.

[0062] As described above, according to the security test system 1 which is one embodiment of the present invention, in order to handle a type of Web page that requires multiple operations to be performed in an appropriate order within the Web page, in addition to the conventional method of obtaining appropriate operation procedures based on predefined rules for what order operations should be performed in what situations, it is possible to create an appropriate operation procedure manual and simulate screen operations using a method that utilizes LLM4 to obtain operation procedures that are close to the content of operations performed by a human.

[0063] In addition, to handle the type of web pages that require operations to be performed in the right order across multiple web pages, LLM4 judges whether or not it is predicted that the behavior of the target web page will change between when it was visited in the past and when it is visited again now, and if it is predicted that a change will occur, it is removed from the grouping and made the target for re-visiting, assuming that the operating procedures are different even for the same URL. This makes it possible to grasp the appropriate operating procedures for target websites3 that include operations that require a sequence, and to perform efficient and effective automatic patrolling.

[0064] The invention made by the present inventor has been specifically described above based on the embodiment, but it goes without saying that the present invention is not limited to the above embodiment and can be modified in various ways without departing from the gist of the invention. Furthermore, the above embodiment has been described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to having all of the configurations described. Furthermore, it is possible to add, delete, or replace part of the configuration of the above embodiment with other configurations.

[0065] In addition, the above-mentioned configurations, functions, processing units, processing means, etc. may be realized in part or in whole by hardware, for example, by designing them as integrated circuits. In addition, the above-mentioned configurations, functions, etc. may be realized in software by a processor interpreting and executing a program that realizes each function. Information such as the program, table, file, etc. that realizes each function can be stored in a recording device such as a memory, a hard disk, or an SSD, or in a recording medium such as an IC card, an SD card, or a DVD.

[0066] In addition, in each of the above figures, the control lines and information lines are shown as those considered necessary for explanation, and do not necessarily show all the control lines and information lines in the actual implementation. In reality, it can be considered that almost all the components are connected to each other. [Industrial Applicability]

[0067] The present invention can be used in a security test system that checks for vulnerabilities in Web applications. [Explanation of symbols]

[0068] 1...Security test system, 2...User terminal, 3...Target website, 4...LLM, 11...inspection management section, 12...patrol processing section, 13...inspection implementation section, 14...patrol target page list, 15...operation procedure manual creation rules, 16...page list, 17...inspection results

Claims

1. A security test system for testing whether or not there is a security vulnerability in a web application, comprising: acquiring and analyzing a web page to be traversed in the web application to be inspected, setting information related to the acquired web page content as a prompt and inputting the information into an LLM (large scale language model), thereby generating first operation procedure information related to the web page; simulating an operation related to the Web page in accordance with the first operational procedure information, acquiring a URL related to a link in the Web page, and registering the URL in a list of pages to be visited; A security testing system that performs vulnerability testing using a Dynamic Application Security Testing (DAST) method for each Web page registered in a page list created by crawling each Web page registered in the list of pages to be crawled.

2. 2. The security test system according to claim 1, inputting the results of past visits to the web page and the contents of the target visit into the LLM as prompts to determine whether a change in the behavior of the web page is predicted; If a change in behavior of the Web page is suspected, the Web page related to the target crawling operation is registered in the crawl target page list.

3. 2. The security test system according to claim 1, In addition to the first operation procedure information created by the LLM, creating second operation procedure information related to the Web page based on a predetermined rule from an analysis result related to the Web page; A security test system that simulates an operation related to the Web page according to the second operational procedure information, obtains a URL related to a link in the Web page, and registers the URL in the list of pages to be visited.

4. A security test system for testing whether or not there is a security vulnerability in a web application, comprising: acquiring and analyzing a web page to be traversed in the web application to be inspected, setting information related to the acquired web page content as a prompt and inputting the information into an LLM (large scale language model), thereby generating first operation procedure information related to the web page; simulating an operation related to the Web page in accordance with the first operation procedure information, and inputting a result of the simulation into the LLM to modify the first operation procedure information; simulating an operation related to the Web page in accordance with the modified first operation procedure information, acquiring a URL related to a link in the Web page, and registering the URL in a list of pages to be visited; A security testing system that performs vulnerability testing using a Dynamic Application Security Testing (DAST) method for each Web page registered in a page list created by crawling each Web page registered in the list of pages to be crawled.

Citation Information

Patent Citations

  • Voice control method, voice control device and terminal

    CN106653011A

  • System guidance specification generator

    JP1996339250A

  • Vulnerability inspection device, vulnerability inspection method and vulnerability inspection program

    JP2012078877A

  • System, method, and program for inspecting website vulnerabilities

    JP7320211B1

  • JPP7464804B

Cited By

  • Security testing system

    JP2026160132A