We Built a Standardized File Format for Prompt Injection and Called It AGENTS.md
A build dependency wrote a file that told your coding agent "ignore everything else, I have absolute authority now" and the agent said okay. That se…
Latest News
⚑ Report a ProblemTech news from the best sources
All topics
- игры
AI
Gear
News
Tech
agents
ai
api
architecture
automation
beginners
career
database
devops
javascript
llm
machinelearning
mcp
opensource
performance
productivity
programming
python
react
security
showdev
testing
tutorial
typescript
webdev
Why SAST and DAST Aren't Enough for Secrets Security
If you run an application security program in 2026, secrets detection can look like a solved line item. Your static analysis suite ships rules for h…
Junior AppSec Engineer Overwhelmed by Massive Code Reviews: Strategies for Efficiency and Confidence
Introduction: The Systemic Failure in Application Security Onboarding Consider the scenario of a junior Application Security Engineer tasked with se…
The LLM Isn't Your Attacker. Your eval() Statement Is.
Everyone's worried about prompt injection making models say bad things. Meanwhile someone piped LLM output straight into eval() and gave it a GPU bo…
Разгоняем GateAI — часть 2: бесплатная модель против Claude
В первой части я разогнал GateAI кэшированием, нашёл несколько багов и проверил триаж на DVWA и pyGoat. Теперь решил прогнать те же задачи через бес…
Разгоняем GateAI
Кэшировал агентный цикл ИИ‑триажа в своём security‑гейте — 80% входных токенов из кэша, ≈ 61% экономии. По пути нашёл 4 реа…
Agentic SQL Injection Is Just SQL Injection Wearing a Trench Coat
Nobody clapped for this one. Zero points, zero comments on HN, and yet CVE-2026-18830 is a better predictor of where agentic AI security is headed i…
Тот самый харнесс который мы заслужили в эпоху, когда безопасность ИИ уже не диковинка
Если запустить любую передовую языковую модель из коробки и попросить ее провести тестирование безопасности агентной системы или собрать актуальный…
OpenAI's New Security Controls Are an Admission, Not an Innovation
An incident happens, a vendor ships a fix, and everyone calls it "proactive security." It isn't. Let's just say that out loud. When a frontier AI la…
CoSnitch Is a Reminder That Your Chatbot Will Tell on You If You Ask Nicely Enough
An AI assistant got talked into describing its own guts to a stranger. That's not a jailbreak curiosity, that's reconnaissance-as-a-service, and it…
OAuth Tokens Were Always the Weak Link. AI Agents Just Made It Worse
Nobody phished anyone in the Vercel or Composio breaches. That's the part worth sitting with for a second, because for the last decade "security awa…
GhostSplice Isn't a Jailbreak, It's a Reminder That LLMs Can't Do Access Control
Split the instruction, split the blame Here's the part that should bother you: nobody had to find a clever new exploit primitive to pull this off. T…
FastAPI accepts a `response_model` on streaming routes and silently ignores it
It builds the model. It registers it in the OpenAPI schema components. It never applies it. Your private fields go out on the wire. Affects FastAPI…
Five findings in FastAPI, and what happened when I reported them
An introduction to a short series. Four of the five are still live in the current release, and each post carries a reproduction you can run. I use F…
A CVSS 10.0 in Your AI Coding Agent Is Just TOCTOU Wearing a Hoodie
A GitHub issue can now potentially exfiltrate your CI secrets, and the tool that let it happen is the same one your team is using to "move faster."…
A text message that runs a command: OS command injection in Gammu SMSD (GHSA-9vjj-v46c-c5qf)
TL;DR What: Gammu SMSD — the daemon behind a huge number of SMS gateways, alerting rigs and 2FA senders — runs an operator-configured hook every tim…
They scoped the customer and forgot the customer's ledger — a High-severity cross-tenant BOLA in Open Food Network
TL;DR What: POST /api/v1/customer_account_transaction in Open Food Network authorized against the class , not the record. It took customer_id straig…
No, Claude Didn't 'Go Rogue.' Someone Gave It Root and Wi-Fi.
When an AI agent breaches a real system, the instinct is to ask "what's wrong with the model?" Wrong question. The right question, the one Anthropic…
GitHub finds your vulnerabilities. Something still has to manage them.
Turn on CodeQL, Dependabot, and Secret Scanning across a handful of repositories and you get exactly what the marketing promised: findings. Lots of…
MLSecOps от кода до продакшена: разные акценты для разных сценариев использования ИИ
В этой статье рассмотрим два сценария использования ИИ в разработке : в первом случае ИИ помогает разработчику писать код, во втором ИИ становится ч…
Звезда в машинном тумане: как ИИ стал оружием, целью и чужим голосом в браузере
AI-ассистенты уже давно стали рабочим инструментом: они помогают писать код, анализировать документы, отвечать на письма, запрашивать данные из друг…
Copilot for Word Will Copy Its Own Poison Into Every Document It Touches
A researcher just disclosed something that should worry anyone using Microsoft 365 Copilot for document generation: Copilot for Word can be manipula…
Adversarial Comments Are Now a Vulnerability Detection Bypass Technique
Your LLM-based vulnerability scanner just cleared a PR with a real, exploitable bug in it. Not because the scanner is dumb. Because someone wrote a…
BloodHound for AI Agents Means We've Officially Given Up Pretending This Is Simple
Two points and zero comments on Hacker News is not a signal of importance. It's a signal that most people haven't figured out yet why this matters.…
Confronting Vault Sprawl And The Risks It Brings
Modern enterprises do not set out to create a maze of credentials, keys, and secrets stores. However, this is the reality most organizations find th…
Running Snyk on Real Legacy Java Code — The Full Unfiltered Results
Numbers are easy to skim. 10 Critical. 99 High. 59 Medium. 20 Low. 188 total. Those numbers appeared in the first article and they're striking — but…
Your AI Coding Assistant Isn't Reading Your Code, It's Mailing It Home
Hook An AI coding CLI that uploads your entire Git history — commit logs, secrets, and all — to a vendor-controlled bucket, and does it through a ch…
Your AI Agent's Memory Is Now an Attack Surface, and Nobody Designed for That
Your AI Agent's Memory Is Now an Attack Surface, and Nobody Designed for That One email. No malware, no exploit chain, no credential theft. Just a h…
Жизненный цикл API-токенов: От генерации до компрометации
JWT не защищает от кражи. Подпись гарантирует, что токен не подделан, но не мешает злоумышленнику использовать его, если тот оказался в чужих руках.…
Сколько Shadow API спрятано в ваших продуктах и как вывести их из тени
Теневые API редко появляются из злого умысла. Чаще это побочный эффект быстрой разработки: временный маршрут для пилота, отдельная точка доступа для…