<?xml version="1.0" encoding="utf-8"?><feed xmlns="http://www.w3.org/2005/Atom" ><generator uri="https://jekyllrb.com/" version="4.3.4">Jekyll</generator><link href="https://aydinnyunus.github.io/feed.xml" rel="self" type="application/atom+xml" /><link href="https://aydinnyunus.github.io/" rel="alternate" type="text/html" /><updated>2026-06-27T22:41:43+00:00</updated><id>https://aydinnyunus.github.io/feed.xml</id><title type="html">Yunus Aydın Blog</title><subtitle>Security researcher exploring cybersecurity, vulnerability research, and responsible disclosure. Discover CVEs, conference talks, and security research.</subtitle><author><name>Yunus Aydın</name></author><entry xml:lang="tr"><title type="html">AI Az Önce Bir Secret Sızdırdı</title><link href="https://aydinnyunus.github.io/2026/06/30/hunting-leaked-secrets-on-github-archive-tr/" rel="alternate" type="text/html" title="AI Az Önce Bir Secret Sızdırdı" /><published>2026-06-30T00:00:00+00:00</published><updated>2026-06-30T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/30/hunting-leaked-secrets-on-github-archive-tr</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[Microsoft, Google, Red Hat, Grafana ve LlamaIndex public GitHub repolarına canlı verified credential pushladı. Pipeline'ı ben yazdım: Gemini 2.5 ile etiketle, regex'e distil et, TruffleHog'u active verification ile koş. 3,830+ verified secret, 1,443 unique repo, %95 rotation.]]></summary></entry><entry xml:lang="en"><title type="html">Your AI Just Leaked a Secret</title><link href="https://aydinnyunus.github.io/2026/06/30/hunting-leaked-secrets-on-github-archive/" rel="alternate" type="text/html" title="Your AI Just Leaked a Secret" /><published>2026-06-30T00:00:00+00:00</published><updated>2026-06-30T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/30/hunting-leaked-secrets-on-github-archive</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[Microsoft, Google, Red Hat, Grafana and LlamaIndex all shipped verified live credentials to public GitHub. I built the pipeline that found them: Gemini 2.5 distilled into regex, then TruffleHog with active verification. 3,830+ verified secrets, 1,443 unique repos, 95% rotation rate.]]></summary></entry><entry xml:lang="tr"><title type="html">Sahte bir iş teklifi aldım. Projedeki npm paketi tam donanımlı bir info-stealer çıktı.</title><link href="https://aydinnyunus.github.io/2026/06/22/fake-job-offer-npm-supply-chain-malware-foxtopia-tr/" rel="alternate" type="text/html" title="Sahte bir iş teklifi aldım. Projedeki npm paketi tam donanımlı bir info-stealer çıktı." /><published>2026-06-22T00:00:00+00:00</published><updated>2026-06-22T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/22/fake-job-offer-npm-supply-chain-malware-foxtopia-tr</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[Freelance NFT staking projesi kılığında bir sosyal mühendislik saldırısı. Zip dosyasındaki typosquat npm paketi (pretie_x1), tarayıcı şifreleri, kripto cüzdanları, SSH anahtarları ve işletim sistemi parolalarını çalan çok aşamalı, AES-şifreli bir info-stealer indirip çalıştırıyor.]]></summary></entry><entry xml:lang="en"><title type="html">I received a fake job offer. The npm package in the project was a full info-stealer.</title><link href="https://aydinnyunus.github.io/2026/06/22/fake-job-offer-npm-supply-chain-malware-foxtopia/" rel="alternate" type="text/html" title="I received a fake job offer. The npm package in the project was a full info-stealer." /><published>2026-06-22T00:00:00+00:00</published><updated>2026-06-22T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/22/fake-job-offer-npm-supply-chain-malware-foxtopia</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[A social engineering attack disguised as a freelance NFT staking project. The zip file contained a typosquatted npm package (pretie_x1) that drops a multi-stage, AES-encrypted info-stealer targeting browser credentials, crypto wallets, SSH keys, and OS login passwords.]]></summary></entry><entry xml:lang="tr"><title type="html">Odysseus: giriş yapmış herhangi bir kullanıcı sunucu genelindeki embedding endpoint’ini ele geçirebiliyor (broken access control + SSRF)</title><link href="https://aydinnyunus.github.io/2026/06/16/odysseus-embedding-endpoint-takeover-tr/" rel="alternate" type="text/html" title="Odysseus: giriş yapmış herhangi bir kullanıcı sunucu genelindeki embedding endpoint’ini ele geçirebiliyor (broken access control + SSRF)" /><published>2026-06-16T00:00:00+00:00</published><updated>2026-06-16T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/16/odysseus-embedding-endpoint-takeover-tr</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[Odysseus'taki POST /api/embeddings/endpoint sadece auth ile korunuyor, admin kontrolü yok. Admin olmayan herhangi bir kullanıcı sunucu genelindeki embedding URL'sini saldırgan kontrolündeki bir hosta yönlendirip tüm kullanıcıların chat, RAG, memory ve vault metnini düz metin olarak dışarı çıkarabiliyor. Aynı handler URL doğrulaması da yapmıyor, yani SSRF de var.]]></summary></entry><entry xml:lang="en"><title type="html">Odysseus: any logged-in user can hijack the server-wide embedding endpoint (broken access control + SSRF)</title><link href="https://aydinnyunus.github.io/2026/06/16/odysseus-embedding-endpoint-takeover/" rel="alternate" type="text/html" title="Odysseus: any logged-in user can hijack the server-wide embedding endpoint (broken access control + SSRF)" /><published>2026-06-16T00:00:00+00:00</published><updated>2026-06-16T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/16/odysseus-embedding-endpoint-takeover</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[POST /api/embeddings/endpoint in Odysseus is auth-gated but not admin-gated. Any non-admin user can repoint the server-wide embedding URL to attacker-controlled hosts, exfiltrating every user's chat, RAG, memory and vault text in plaintext. The same handler does no URL validation, so it doubles as SSRF.]]></summary></entry><entry xml:lang="tr"><title type="html">NLTK collocations’da eval() üzerinden komut enjeksiyonu</title><link href="https://aydinnyunus.github.io/2026/06/07/command-injection-nltk-collocations-eval-tr/" rel="alternate" type="text/html" title="NLTK collocations’da eval() üzerinden komut enjeksiyonu" /><published>2026-06-07T00:00:00+00:00</published><updated>2026-06-07T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/07/command-injection-nltk-collocations-eval-tr</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[NLTK collocations.py'de sys.argv üzerinden eval() ile komut enjeksiyonu zafiyeti. Keyfi Python kodu çalıştırma gösterildi; ancak istismar lokal CLI erişimi gerektiriyor.]]></summary></entry><entry xml:lang="en"><title type="html">Command injection in NLTK collocations via eval()</title><link href="https://aydinnyunus.github.io/2026/06/07/command-injection-nltk-collocations-eval/" rel="alternate" type="text/html" title="Command injection in NLTK collocations via eval()" /><published>2026-06-07T00:00:00+00:00</published><updated>2026-06-07T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/06/07/command-injection-nltk-collocations-eval</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[Command injection vulnerability in NLTK collocations.py via eval() on sys.argv. Arbitrary Python code execution demonstrated, though exploitation requires local CLI access.]]></summary></entry><entry xml:lang="tr"><title type="html">CVE-2026-5728: LollMS chat image upload’unda Content-Type spoofing</title><link href="https://aydinnyunus.github.io/2026/05/03/content-type-spoofing-lollms-chat-image-cve-2026-5728-tr/" rel="alternate" type="text/html" title="CVE-2026-5728: LollMS chat image upload’unda Content-Type spoofing" /><published>2026-05-03T00:00:00+00:00</published><updated>2026-05-03T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/05/03/content-type-spoofing-lollms-chat-image-cve-2026-5728-tr</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[CVE-2026-5728: LollMS /api/upload/chat_image endpoint'i sadece istemcinin gönderdiği Content-Type başlığına güveniyor; doğrulanmış kullanıcılar PNG ya da JPEG kılığında görsel olmayan dosyalar yükleyebiliyor.]]></summary></entry><entry xml:lang="en"><title type="html">CVE-2026-5728: Content-Type spoofing on LollMS chat image upload</title><link href="https://aydinnyunus.github.io/2026/05/03/content-type-spoofing-lollms-chat-image-cve-2026-5728/" rel="alternate" type="text/html" title="CVE-2026-5728: Content-Type spoofing on LollMS chat image upload" /><published>2026-05-03T00:00:00+00:00</published><updated>2026-05-03T00:00:00+00:00</updated><id>https://aydinnyunus.github.io/2026/05/03/content-type-spoofing-lollms-chat-image-cve-2026-5728</id><author><name>Yunus Aydın</name></author><summary type="html"><![CDATA[CVE-2026-5728: LollMS /api/upload/chat_image trusts the client Content-Type header only, so authenticated users can upload non-images disguised as PNG or JPEG.]]></summary></entry></feed>