All writing

WEBCRAWLER & SQLI-FINDER REWORK

These two projects were part of my 2025 web security products, which were functional but neither efficient nor properly structured.

Take SQLi-Finder, for example. It was purely written as a dork-finder, where you would enter a parameter such as id or product and check whether adding an apostrophe to the end of the URL changed the website's contents. Even though this was a red flag that could imply an SQL injection vulnerability, it could also stem from other reasons, and relying purely on such changes could mistake ordinary website behavior for an actual vulnerability. Because the tool also lacked security measures and authorization checks, I kept it private for nine months.

WebCrawler, on the other hand, was able to list each page of a website using a depth-first search (DFS) structure. However, it was only a reporting mechanism, and other than that, it would crash every time the headless browser was closed, displaying unnecessary error messages in the terminal.

Having this in mind all along, after finalizing my submission draft for the Gambling And Its Digital Distribution paper, I started working.

SQLi-Finder's old architecture encouraged broad public-web scanning rather than explicit-scope testing. Dynamic pages could also change between requests, so page-source differences alone produced noisy false positives. Instead of relying solely on those differences, I introduced comparisons of database-error signatures and response behavior, allowing the tool to distinguish stronger indicators from changes that might have nothing to do with SQL injection.

Adding eight CLI options and five result states (likely, possible, not_detected, skipped, and error), I turned SQLi-Finder into a specialized SQL injection indicator scanner. With additional features such as checking multiple URLs at a time, reading targets from a file, configurable request delays, timeouts, and explicit authorization confirmation, the tool became worthy of publishing. It no longer needed to rely on search-engine dorks or browser automation for its core functionality, and its findings were now classified as indicators requiring verification rather than definitive proof of a vulnerability.

WebCrawler, on the other hand, was indeed functional, but I added classifications for admin pages, potential SQL injection candidates, and XSS-related input surfaces. I also improved its handling of headless-browser availability, allowing it to continue with ordinary HTTP requests when Selenium or Chrome was unavailable, rather than simply failing. Checkpointing and resumable crawling were added as well, allowing interrupted scans to continue without starting everything from scratch.

I intentionally left active vulnerability testing out of WebCrawler, so it became a cyber-reconnaissance tool rather than another vulnerability scanner. It identifies potentially interesting pages, parameters, and input fields for further investigation without attempting to exploit them.

Both projects had started as functional but limited tools in 2025. The rework was not about adding as many features as possible. It was about addressing the limitations I had left unresolved and making the tools more structured, practical, and responsible to use.

All writing