This repository contains a collection of intentionally vulnerable Model Context Protocol (MCP) server implementations for security training, covering common failure modes including path traversal, indirect prompt injection, eval-based RCE, namespace typosquatting, and secrets exposure. Each server includes documentation on exploitation techniques, supporting hands-on training for pentesters learning AI agent security and MCP integration risks. Intended for isolated lab environments only.
reNgine is an automated reconnaissance framework for web applications focused on highly configurable, streamlined recon processes via Engines, recon data correlation, and organization. It supports continuous monitoring and is backed by a database.
Betterment disclosed that attackers used social engineering to access some customer data, including names and contact details, via third-party marketing and operations platforms, and then pushed a fake in‑app crypto message promising to triple deposits sent to a malicious wallet. The firm says no accounts or credentials were compromised. It is investigating the incident with external cybersecurity support.
The Russian Fancy Bear APT has been carrying out a low-complexity, spear-phishing campaign targeting governments or organizations of strategic value to Russia. The phishing emails contain a link to a document that the target believes is relevant, which then redirects them to a phishing page for an email or VPN provider. While Fancy Bear is better known for more sophisticated malware campaigns, this campaign may have been chosen as a low-cost effort to gain access to notable organizations.
AWS Service-Linked Roles (SLRs) are a special class of IAM roles managed by AWS and used to bootstrap services or resources within services that depend on other services. SLRs are largely undocumented, so the author of this post used the undocumented iamv2 API to enumerate the trust policies and IAM policies of the SLRs used by as many services as they could find. They discovered several “code smells” in the IAM policies, such as a lack of confused-deputy protection, resource policies that use wildcards, tag-based policies, or name-based policies.
Europol-coordinated raids across Spain arrested 34 members of the Black Axe cybercrime organization and seized €185,000 in assets linked to romance scams, phishing, and BEC fraud, resulting in nearly €6 million in losses.
Ireland recalled nearly 13,000 passports issued between December 23 and January 6 after a software update omitted the 'IRL' country code from the machine-readable zone, rendering documents non-compliant with ICAO standards and potentially unusable at automated border gates.
Infoblox researchers exposed service providers, including Penguin Account Store and UWORK, that supply Chinese-speaking crime syndicates with turnkey pig butchering-as-a-service (PBaaS) infrastructure, offering fraud kits, stolen credentials, CRM platforms, and complete scam packages starting at $50.
Huntress observed Chinese-speaking threat actors exploiting three VMware ESXi zero-days (CVE-2025-22224, CVE-2025-22225, and CVE-2025-22226) to escape virtual machine isolation and gain full hypervisor control. Evidence suggests the exploit toolkit was developed as early as February 2024, over a year before Broadcom's March 2025 disclosure. The multi-stage attack chain used a compromised SonicWall VPN for initial access and deployed the VSOCKpuppet backdoor, which communicates over VSOCK to bypass traditional network monitoring entirely. Organizations should prioritize patching these vulnerabilities and monitor for unusual VMCI/VSOCK activity between guest VMs and hypervisors.
The latest version of the infamous BreachForums suffered a data breach that leaked a database of user accounts. The breached data includes display names, registration dates, IP addresses, and other internal information. Many of the breached users' IPs route to loop back, but over 70k records contain public IPs.
Malwarebytes raised concerns about a trove of Instagram user information being sold on the dark web. The breached data includes usernames, emails, phone numbers, and partial location data. Instagram has stated that there was no data breach, but rather an issue that allowed external parties to send unrequested password reset emails.
ORM Leak vulnerabilities extend beyond susceptible ORMs such as Django and Prisma to any application that implements robust filtering capabilities. This article details practical examples, including Harbor CVE-2025-30086 (Beego ORM) and authentication bypasses via Prisma operator injection via type coercion, and discusses multiple techniques for bypassing naive protections, including exploiting Beego's broken expression parsing, where non-relational fields in filter expressions are overwritten by subsequent fields. Semgrep detection rules for Django, Prisma, Beego, and Entity Framework ORMs have been published to help identify these commonly overlooked vulnerabilities.
Mailpit is a lightweight email testing tool used in development that recently had two notable security issues uncovered: a server-side request forgery (SSRF) and a cross-site WebSocket hijacking (CSWSH) flaw. The SSRF bug allowed attackers to abuse a proxy endpoint to access internal services such as cloud metadata, while the WebSocket issue allowed any website to connect and stream all captured emails without authentication. The maintainer quickly fixed both issues in new releases and coordinated CVEs, showing how respectful communication, GitHub Security Advisories, and low-ego collaboration between researchers and developers lead to fast, effective vulnerability remediation.
GitHub ranks as the second most used CI/CD tool, but it also exposes a large attack surface to attackers. GitHub Actions can unintentionally expose secrets or overprovision tokens. Actions are vulnerable to supply chain compromise via poisoned dependencies and injection attacks.
Tailsnitch is a security auditing tool for Tailscale configurations that scans tailnets for 50+ misconfigurations, overly permissive access controls, and security best practice violations across seven categories, including ACLs, authentication keys, and network exposure. The tool supports severity filtering, interactive remediation mode, JSON export for CI/CD integration, and SOC 2 evidence generation with Common Criteria control mappings. Checks range from critical issues like default allow-all policies and overly broad tagOwners to informational items covering logging and DNS settings.
Gixy-Next (Gixy) is an open-source NGINX configuration security scanner and hardening tool that statically analyzes nginx.conf files to detect security misconfigurations, hardening gaps, and common performance pitfalls before they reach production. It is an actively maintained fork of Yandex's Gixy.
Iranian threat actor MuddyWater has been attributed to spear-phishing campaigns targeting diplomatic, maritime, financial, and telecom entities in the Middle East using a new Rust-based implant called RustyWater (also known as Archer RAT/RUSTRIC). The malware, delivered via malicious Word documents with VBA macros, establishes registry persistence and supports asynchronous C2 communication for file operations and command execution. This marks MuddyWater's continued evolution away from PowerShell and VBS loaders toward more structured, modular RAT capabilities.
Europol and Spanish police have dismantled a major Black Axe cell in Spain, arresting 34 suspects across Seville, Madrid, Málaga, and Barcelona for cyber-enabled fraud and violent organized crime. Authorities link the Nigerian-origin syndicate to scams like business email compromise and romance fraud, with losses over €5.93 million, and report cash and bank account seizures alongside prior global crackdowns on the group.
The primary cybersecurity question of 2026 will be whether an attacker's or defender's AI is more effective. Agentic security will encourage teams to build products rather than buy them, devalue junior employees, and lead to workplace upheaval. AI will also improve the efficacy of common security program weak points, such as secure code training and asset management.
Anthropic confirmed that a viral screenshot claiming Claude permanently banned a user and reported them to authorities is fake, noting the recurring hoax doesn't match any actual system messages.
Microsoft is testing a new RemoveMicrosoftCopilotApp group policy in Windows 11 Insider builds that allows IT administrators to uninstall the Copilot app on managed devices via Intune or SCCM.
Instagram acknowledges that an external party exploited a flaw to trigger mass password reset emails, but insists there was no breach and that accounts remain secure, advising users to ignore the notifications
AWS-native tools protect what's inside AWS. But modern attacks move across identity, SaaS, on-prem, and other clouds, before and after they ever touch AWS.
Illinois' Department of Human Services left an internal mapping site publicly accessible for more than four years, exposing data on over 700,000 people receiving state benefits. The leak revealed addresses, case numbers, demographic details for Medicaid and Medicare Savings Program recipients, and names and case information for tens of thousands of individuals receiving rehabilitation services.
The European Space Agency faces a new breach after Scattered/Shiny Lapsus$ Hunters claimed to have stolen 500 GB of highly sensitive operational, spacecraft, and contractor data from ESA systems, including information on missions involving partners such as SpaceX and Airbus. The attackers say their access remains open. The ESA has referred the case for a criminal inquiry amid a history of prior security incidents dating back to 2011.
The popular jsPDF library is vulnerable to a local file inclusion and path traversal attack that can allow attackers to steal secrets. The flaw allows attackers to pass unsanitized paths directly to the loadFile function or any functions that call it. The vulnerability was fixed in version 4.0.0, but it may not be comprehensive on Node 20 and earlier.
Following recent npm supply chain attacks, GitHub's hardening guidance centers on three controls: adopting Trusted Publishing Framework (OIDC-based short-lived credentials instead of long-lived tokens), enforcing 2FA for all publishing operations, and using WebAuthn/passkeys instead of TOTP for phishing-resistant authentication. npm deprecated new TOTP setups in October and disabled legacy tokens in November, requiring maintainers to use granular access tokens with expiration for CI/CD automation. Maintainers should configure trusted publisher relationships in package settings, enable 2FA across all sensitive operations, and enroll WebAuthn authenticators while maintaining repository access controls.
2026.01.09·
securitylabs.datadoghq.com·13 min read·TLDR InfoSec
A critical pre-auth remote code execution flaw (CVE-2025-52691) in SmarterTools SmarterMail, a business email and collaboration server for Windows and Linux, arises from an unauthenticated file upload endpoint that accepts attacker-controlled parameters. It includes a GUID used to construct file paths, enabling arbitrary file writes via path traversal and by crafting a multipart/form-data request with specific context and contextData values. With a malicious file, an attacker can place a web shell in the webroot and gain full remote code execution.
Logs are an essential tool for defenders to monitor for breaches and investigate incidents. However, they can also be a valuable enumeration tool for attackers. This article provides a comprehensive overview of how attackers can abuse logs for enumeration in AWS, Azure, and GCP. Security teams should treat logs with care and enforce the principle of least privilege where possible.
You don't want to deal with data visibility nightmares to understand if a threat is real. Sumo Logic unifies AWS, GCP, Azure, and 450+ other tools into a single platform - powered by multi-agent AI [13] that surfaces root causes, prioritizes key events, and guides analysts through resolution. Read the Gartner Critical Capabilities report [14] to learn how Sumo Logic compares to other SIEM competitors.
Blackbird.AI provides an AI-powered narrative intelligence platform that helps organizations detect, analyze, and respond to harmful or misleading online narratives and influence campaigns that threaten reputation, operations, and security.
The PQC Network Scanner is a quantum-focused network assessment tool that scans TLS/SSL certificates across environments to identify systems vulnerable to future quantum attacks.
Go's go.sum file is not a lockfile and has no effect on version resolution—it's purely a local cache for the Go Checksum Database to ensure cryptographic integrity. Developers should use go.mod for dependency analysis, which, since Go 1.17, lists all transitive dependencies with exact versions and functions as both a manifest and lockfile. This design eliminates diamond dependency conflicts and prevents automatic upgrades to potentially compromised dependency versions.
Analysis of 125,183 Linux kernel bugs with Fixes: tags revealed an average bug lifetime of 2.1 years, with race conditions averaging 5.1 years, and the longest-lived bug persisted for 20.7 years. Detection rates improved dramatically, from 0% bugs found within one year in 2010 to 69% in 2022, driven by tools such as Syzkaller and kernel sanitizers. The VulnBERT model achieved 92.2% recall and a 1.2% false-positive rate by combining CodeBERT embeddings with 51 handcrafted features targeting patterns such as unbalanced refcounts and missing NULL checks.
Security researchers at Radware show how a new “ZombieAgent” attack revives a previously blocked ChatGPT data‑exfiltration technique by tweaking prompts to bypass URL restrictions and leak information character by character. Indirect prompt injection and weak separation between user instructions and embedded content keep LLM agents vulnerable, forcing vendors into an endless patch-and-bypass cycle rather than a fundamental fix.
Google filed a lawsuit against SerpApi for circumventing security measures through bot networks, cloaking techniques, and fake crawler identities to scrape and resell copyrighted content from search results.
Gulshan Management Services, operator of Handi Plus and Handi Stop gas stations, disclosed a September 2025 breach exposing SSNs, financial data, and government IDs of 377,000+ individuals.
Cybersecurity researchers reveal 11 critical vulnerabilities in Coolify, a popular self‑hosting platform, that enable authenticated low‑privileged users to achieve root command execution, escape containers, and even obtain the server's root SSH key, potentially compromising entire infrastructures.
Today's email security stops at the perimeter. Once an attacker slips past your inbound detections, they have free reign over years of sensitive emails, files, and internal comms.
CVE-2026-0625, a critical command injection vulnerability in legacy D-Link DSL routers' dnscfg.cgi endpoint, has been actively exploited in the wild since November 2025, targeting end-of-life models including DSL-2740R, DSL-2640B, DSL-2780B, and DSL-526B. The unauthenticated remote code execution flaw enables attackers to hijack DNS settings without credentials, allowing them to redirect traffic and persistently compromise all downstream devices behind the router. Organizations must immediately retire affected unpatchable devices and upgrade to actively supported models, as the vulnerability leverages the same DNS configuration mechanism used in previous large-scale DNSChanger campaigns.
Australian rental car insurance provider Prosura suffered a data breach that impacted an undisclosed number of customers. The leaked data may include names, email addresses, phone numbers, country of residence, travel details, invoicing and pricing data, and policy dates. The threat actor also directly contacted impacted customers, alleging that the company has been ignoring them.
Ledger has informed some customers who made purchases on ledger.com that their personal data was compromised as part of a breach impacting the third-party payment processor Global-e. Ledger and Global-e have confirmed that no financial information or wallet seed phrases were leaked as part of this breach. Global-e is also notifying other impacted companies that utilize its services.
The 2025 macOS threat landscape was characterized by a surge in sophisticated infostealers and state-sponsored backdoors that increasingly leverage "Malware-as-a-Service" models and social engineering vectors such as "ClickFix" and fake job interviews. Technical analysis reveals that attackers are bypassing Gatekeeper and static analysis by utilizing signed or notarized binaries and executing multi-stage payloads written in Go, Python, and AppleScript directly in memory. To combat these evolving threats, security professionals must move beyond static signatures to implement behavioral monitoring of process, file, and network activity while educating users on the dangers of executing terminal commands from untrusted sources.
Cronometer's Android app uses Flutter's own TLS stack, which ignores system proxies and bakes certificate pinning into libflutter.so, making traditional MITM approaches fail. By combining Frida on a rooted Pixel with custom hooks, it becomes possible to detect and patch Flutter's TLS verification, observe DNS lookups and socket connections, and then transparently redirect selected hosts through a proxy like mitmproxy. Injecting an HTTP CONNECT tunnel at the syscall level lets HTTPS traffic flow through the proxy without modifying the APK.