Posts

Oops, No Victims: The Largest Supply Chain Attack Stole 5 Cents

Image
The Biggest NPM Supply Chain Attack What is a Supply Chain Attack? A supply chain attack occurs when attackers target trusted third-party components, such as libraries or registries, instead of attacking users directly. By injecting malicious code at the source, they can spread it to all downstream users. These attacks are dangerous because updates happen automatically in build pipelines, making detection harder. A small modification in a common dependency can silently compromise thousands of projects. Defenses require strong authentication, artifact signing, reproducible builds, and active monitoring of supply chain integrity. Introduction On September 8, 2025, the npm ecosystem faced one of its largest compromises. A maintainer’s account was hijacked, and malicious versions of popular packages were published. Since npm packages are used globally in countless projects, the exposure was immediate and severe. Although the financial damage was limited, the operational dis...

File Doppelgängers: The World of Polyglot Files

Image
Unlocking the Secrets of Polyglot Files: When One File Speaks Many Languages Introduction In the world of cybersecurity, appearances can be deceiving. A file might look like an innocent image, but behind the scenes, it could also be a ZIP archive, a PDF, or even an executable. Such files are called polyglots — single files that are valid under multiple file formats at the same time. Polyglots are fascinating from a technical standpoint, but they also pose serious security risks. Attackers often use them to bypass filters, sneak past antivirus solutions, or trick unsuspecting users into executing malicious code. What Are Polyglot Files? A polyglot file is crafted in such a way that it conforms to the specifications of two (or more) different file types simultaneously. Example: A file that is both a JPEG image and a ZIP archive . When opened in an image viewer → You see a normal picture. When extracted with a ZIP tool → You find hidden file...

Sometime slash might be Unicode Hiragana

Image
Cybercriminals Exploit Unicode Trick to Mimic Booking.com in Phishing Scam Threat actors have launched a new phishing campaign targeting Booking.com users , leveraging Unicode characters to disguise malicious links as legitimate ones. This attack shows how easily scammers can trick people with subtle visual deceptions. How the Attack Works The campaign uses the Japanese hiragana character “ん” (Unicode U+3093) inside phishing URLs. On some systems and fonts, this symbol looks like a forward slash (/) or part of a subdirectory, making the fake URL appear genuine. For example, a phishing email may display the link (appears safe but is deceptive): https://admin.booking.com/hotel/hoteladmin/... But the actual malicious hyperlink is different (neutralized below for safety): https://account.booking.comんdetailんrestric-access.www-account-booking[.]com/en/ At first glance, it looks like a real Booking.com page. In reality, the true registered domain is “www-account-booking[.]co...

Identicon

Image
The Mystery of Identicons: Turning Data into Unique Visual Avatars Introduction If you’ve ever signed up for a new website and seen a colorful, pixelated avatar appear next to your name — without uploading a profile picture — you’ve probably met an Identicon . Identicons are unique, algorithmically generated images based on a piece of text, usually a username, email, or IP address. They are used for visual identification and uniqueness without revealing personal information. Where Do We See Identicons? GitHub: User avatars when no profile picture is uploaded. WordPress: Default avatars for comments. StackOverflow: Generated icons for users without a custom profile picture. How Do Identicons Work? Hashing the Input – The text (e.g., "kiyotaka ayanokōji") is passed through a hash function like SHA1 or MD5, producing a fixed-length hexadecimal value. Mapping Bits to a Grid – The hash is split into smaller parts, each controlling ...

Retrieval Augmented Generation

Image
Mastering RAG: The Future of Fact-Aware AI  “An AI that thinks like L, but researches like Sherlock.” – That’s the power of RAG. What is RAG? RAG stands for Retrieval-Augmented Generation — an advanced AI technique that combines a retriever (like a search engine) with a generator (like GPT). Traditional AI only replies based on training. RAG goes further — it retrieves live facts from external documents before generating an answer. Why RAG Was Introduced Most LLMs are like students who read thousands of books last year — but can’t learn new things today. RAG fixes this by: Reducing hallucinations (wrong facts) Adding real-time knowledge to AI responses Combining search + generation into one smart process What RAG Replaces Before RAG, you needed to: Manually feed facts into prompts Use Google + ChatGPT separately Retrain your models constantly RAG solves this with one pipeline that searches, reads, and generates together. RA...

Convolution Neural Network

Image
What is a Convolutional Neural Network (CNN)?      A Convolutional Neural Network (CNN) is a type of deep learning model specifically designed to process visual data like images and videos. It works by automatically learning patterns such as edges, textures, shapes, or objects from images without needing manual feature engineering. CNNs are inspired by the way the human visual cortex works and are widely used in computer vision tasks like image classification, object detection, face recognition, and more. CNNs reduce the complexity of image data using a structure of layers that include convolution layers, pooling layers, and fully connected layers. Each layer plays a unique role in extracting, simplifying, and interpreting the visual features. By stacking multiple such layers, CNNs can identify complex patterns and even recognize complete objects. One major advantage is that CNNs learn spatial hierarchies—starting from small patterns like edges to entire objects. CNNs ha...

AI Agents Assemble: The Future Just Got Smarter!

Image
AI Agent 🤖             An  AI Agent  is a computer program that can think and make decisions like a human. It works by observing its surroundings, deciding what to do, and then taking action. These agents are designed to solve problems, complete tasks, and even learn from their mistakes. They don’t need someone to tell them what to do every time they can work on their own. AI Agents use logic, planning, and sometimes learning to get better over time. They can respond to changes, make choices, and keep improving. In simple words, an AI Agent is a smart system that can act on its own using artificial intelligence. It behaves in a way that helps it reach its goals effectively. 1. key Features of an AI Agent:              An AI is an smart software program, that can think, learn and take actions on its own, the key features of an AI Agents includes Reasoning, Acting, Observing, Planning, Col...

Flask Cookie

Image
A cookie is a small piece of data stored on the user’s browser by a website. It helps websites remember user preferences, authentication status, and other session-related data. Key features of ccookies: ✅ Stored on the client-side (browser). ✅ Sent with each request to the server. ✅ Used for sessions, authentication, tracking, and personalization. ✅ Can have attributes like expiration time, HTTPOnly, Secure, and SameSite for security. 🏗 What is a Flask Cookie? A Flask cookie is a way to store data on the user’s browser using Flask’s set_cookie() method. Flask allows developers to set, read, and delete cookies easily. How Flask Uses Cookies: 1️⃣ Setting a Cookie: resp.set_cookie('key', 'value') 2️⃣ Getting a Cookie: request.cookies.get('key') 3️⃣ Deleting a Cookie: resp.set_cookie('key', '', expires=0) from flask import Flask, render_template, request, url_for, redirect, make_response, flash, session import rand...