Featured · Product Updates
The Brain: giving security agents knowledge of your environment
The Brain is Cantina's shared context and memory layer for AI security agents.
Cantina 5 min read
Read article
Cantina Blog
Vulnerability Research, practical guides, and updates from the people building Cantina.
83 stories
All articles
25–36 of 83 articles
Category
-
Vulnerability Research
PHP2Shell: What’s affected and how to patch
Connect breaking threat intelligence to your SBOM, alert the right owners, and carry the response through remediation and verification.
Cantina 3 min read
-
Vulnerability Research
Apex Found Two Authentication Bypasses in wolfSSL
Apex found two configuration-specific authentication bypasses in wolfSSL. Both were fixed in version 5.9.4.
Cantina Security Research 7 min read
-
Product Updates
What We Shipped: Introducing Apex Flash-1
Meet Apex Flash-1, plus new ways to create Clarion agents, prioritize Apex findings, and investigate with context from the Brain.
Cantina 4 min read
-
Vulnerability Research
Storybook Could Let Someone on Your Network Change Your Code
Apex found a Storybook development-server flaw that could let someone on the same network inject code into a developer's story files.
Cantina Security Research 8 min read
-
Guides
FedRAMP’s new rules: closing the gap between finding and fixing
How recurring Apex scans, scoped pentesting, and fix reviews keep security testing moving as code changes and support FedRAMP vulnerability programs.
Cantina 6 min read
-
Vulnerability Research
Apex Found a 15-Year-Old, High-Severity Bug in XZ Utils
A failed memory allocation left XZ Utils remembering a buffer that no longer existed. Reusing the decoder could crash the application.
Cantina Security Research 9 min read
-
Product Updates
Introducing Apex Flash-1
Cantina and Yeta Labs release Apex Flash-1, an open-weight cybersecurity model built on GLM-5.3-Flash. Explore its training, results, and model downloads.
Cantina 1 min read
-
Vulnerability Research
How Empty Fields Broke OpenEMR's Patient Data Boundary
Empty fields exposed a patient's consent document while a denied request still changed who could read a protected file.
Cantina Security Research 9 min read
-
Product Updates
The Brain: giving security agents knowledge of your environment
The Brain connects identities, assets and investigation history for Cantina's security agents. See demo investigations and a 40-alert observational comparison.
Cantina 5 min read
-
Company News
Robinhood Chain launches a bug bounty on Cantina
Robinhood Chain's public bug bounty is live on Cantina, with rewards up to $1,000,000 for eligible findings across smart contracts and public applications.
Cantina 2 min read
-
Product Updates
What We Shipped: Claude Code Credential-Harvesting Detection
Clarion now flags potential credential harvesting in Claude Code, with new integrations, Apex finding-review updates, and investigation tools.
Cantina 4 min read
-
Vulnerability Research
Apex Returns to CompCert: A Bug Added After Six CPU-Years of Fuzzing
Apex returned to CompCert and found a bug in code added after its famous fuzzing campaign. A 32-bit choice became a 64-bit jump-table index.
Cantina 9 min read
-
Vulnerability Research
How Clarion brings production context to Dependabot triage
Clarion connects Dependabot alerts to code and production context, makes missing evidence visible, and carries assessments into remediation and reporting.
Cantina 5 min read
-
Product Updates
Get a free FHIR vulnerability scan before the 2027 CMS deadline
Prepare FHIR APIs for CMS-0057-F requirements due primarily January 1, 2027. See five Pathling findings and apply for a free vulnerability scan.
Cantina 3 min read
-
Company News
Cantina joins OpenAI’s call for collective cyber defense
Cantina has signed OpenAI’s call for collective cyber defense, supporting wider access to advanced security capabilities for organizations of every size.
Cantina 2 min read
-
Guides
Creating a Security Exposure Agent
Build a web exposure analysis agent with Cloudflare Workers. Follow the scans, evidence and false positives that show why HTTP 200 is not proof of exposure.
Jimmy Vo 18 min read
-
Product Updates
Claude Code Telemetry in Clarion | August 26 Changelog
Clarion adds Claude Code and MCP server visibility, human approvals and exposure sweeps. Apex gains Audit Setup, CLI updates and clear scan controls.
Cantina 5 min read
-
Vulnerability Research
How Three Numbers Broke Seagate's Memory Bounds
Apex traced three Seagate openSeaChest memory bugs to an overflowing allocation, an inclusive loop, and an unchecked device index.
Cantina 4 min read
-
Vulnerability Research
How We Found Three Bugs in a Compiler Proven Correct
CompCert is a formally verified C compiler for high-assurance software. Its proof held. The bugs lived at the edges.
Cantina 20 min read
-
Vulnerability Research
Two Node.js bugs you should know about
Two Node.js vulnerabilities can replay SQLite writes or crash workers through oversized DNS responses. See the affected paths, fixed releases, and mitigations.
Cantina 7 min read
-
Vulnerability Research
One Emoji Is Enough to Crash Ruby
A single 4-byte emoji overflows a fixed stack buffer in Prism, Ruby's default parser since 3.4. What breaks, why the safety check causes it, and the fix.
Cantina 10 min read
-
Company News
Cantina Launches Clarion With $16.5M in Total Funding
Cantina launches Clarion alongside $8M in new funding led by Framework Ventures, bringing total funding to $16.5M.
Cantina 7 min read
-
Product Updates
Every agent will have a grok build moment. this Is how you handle one
SpaceXAI's Grok Build coding agent was uploading unnecessary repository data; the company's transparent response set industry standards.
Cantina 5 min read
-
Vulnerability Research
What four RabbitMQ advisories reveal about broker security boundaries
Analysis of four RabbitMQ security vulnerabilities affecting OAuth permissions, topic access, loopback enforcement, and Direct Reply-To bindings.
Cantina 7 min read
-
Vulnerability Research
Three trust boundary CVEs in Node.js
Security researchers disclose three distinct vulnerabilities affecting Node.js hostname and TLS stack layers, each enabling authentication bypasses through trust boundary violations.
Cantina 13 min read
-
Company News
Cantina joins Anthropic's cyber verification program
Cantina gains access to Anthropic's CVP to enhance AppSec agent capabilities for vulnerability verification and resolution.
Cantina 5 min read
-
Guides
Deepfakes in 2026: how they evolved, how to spot them, and where they are headed
A comprehensive analysis of deepfake fraud trends, detection methods, and defensive strategies as of 2026.
Cantina 7 min read
-
Vulnerability Research
TeamPCP: the hacker group that turned developer trust into an attack surface
Analysis of how TeamPCP compromised the Trivy security scanner and launched a multi-ecosystem supply chain attack affecting major software delivery infrastructure.
Cantina 12 min read
-
Company News
Cantina CEO Hari Mulackal on the Fable pull: the only clock you control Is how fast you fix
Cantina's CEO examines how AI model capabilities for finding vulnerabilities have shifted security from access control to remediation speed.
Cantina 4 min read
-
Vulnerability Research
TrustLaunder: how a healthcare FHIR server turned attacker URLs into trusted patient data
A vulnerability in Pathling Server versions 1.2.0 and earlier allowed attackers to bypass security allowlists by reclassifying untrusted data as local after fetching it from attacker-controlled URLs.
Cantina 6 min read
-
Vulnerability Research
After TrustLaunder: Apex's other four pathling findings
Four additional high-severity vulnerabilities discovered in Pathling FHIR analytics server beyond the TrustLaunder findings.
Cantina 10 min read
-
Vulnerability Research
Swift Crypto X-Wing HPKE overread
A memory-safety vulnerability in Apple's swift-crypto library allows out-of-bounds reads when processing malformed X-Wing HPKE encapsulated keys.
Cantina 14 min read
-
Vulnerability Research
Race condition in Ruby core
Cantina's Apex agent discovered CVE-2026-46727, a use-after-free race condition in Ruby 4.0.0-4.0.4's DNS resolver that can crash processes.
Cantina 7 min read
-
Vulnerability Research
When AI writes half your code, shift-left becomes a volume problem
AI-assisted development creates security triage bottlenecks that faster scanners cannot solve; routing findings to service owners with drafted fixes is the actual scaling lever.
Cantina 7 min read
-
Vulnerability Research
Where AI-generated code quietly fails: A field note on false negatives
AI-generated code often contains authorization flaws that static analysis misses because the vulnerabilities exist across multiple functions rather than within individual code blocks.
Cantina 9 min read
-
Vulnerability Research
How one VS code extension took down ~3,800 GitHub internal repos
A poisoned VS Code extension compromised a GitHub employee device, exposing 3,800 internal repos in a broader supply chain attack campaign.
Cantina 12 min read
-
Guides
Translating AI agent policy into runtime control: A working map
Governance requires mapping policy clauses to runtime checks and audit evidence, not just writing policies.
Cantina 4 min read
-
Guides
Cantina x Swif: iOS 26.5 security guide, 2026's biggest Apple patch cycle
Analysis of Apple's May 2026 security patch cycle that closed 80+ vulnerabilities, including AI-credited findings from both Cantina and Anthropic.
Cantina 12 min read
-
Vulnerability Research
What four CVEs in fourteen days tell Us about AI infrastructure risk
Four production-critical AI infrastructure vulnerabilities reveal systemic gaps in how organizations manage interconnected security risks across their AI stack.
Cantina 4 min read
-
Vulnerability Research
When a linux networking flaw lets a local foothold turn into root across major distros
Analysis of Dirty Frag, a Linux privilege escalation vulnerability affecting major distributions through networking subsystem flaws.
Cantina 5 min read
-
Vulnerability Research
The FBI's cargo-theft alert shows how load-board fraud becomes freight loss
Cyber-enabled cargo theft reached $725M in losses during 2025, exploiting compromised dispatch workflows and freight software to orchestrate sophisticated theft schemes.
Cantina 5 min read
-
Vulnerability Research
Apex found a 13-year-old bug in WebKit. Apple patched it yesterday
Cantina's autonomous agent Apex discovered a memory safety vulnerability that persisted in WebKit for 13 years, along with two CSP bypasses, all patched in iOS 26.5.
Cantina 19 min read
-
Vulnerability Research
AI agent governance Is moving from policy to runtime control
Governance of AI agents is shifting from policy frameworks to real-time runtime controls as regulatory bodies establish standards and minimum security requirements.
Cantina 5 min read
-
Vulnerability Research
Federal PLC exploitation Is turning exposed controllers into plant disruption
Federal agencies warn of active exploitation of internet-connected PLC controllers across U.S. critical infrastructure, creating operational trust and safety concerns.
Cantina 3 min read
-
Vulnerability Research
When prompt injection turns an AI agent framework into host-level code execution
Prompt injection becomes a critical security issue when AI agent frameworks allow model-controlled input to reach file operations or code execution.
Cantina 3 min read
-
Vulnerability Research
Authorization bypass in Spring security 7: XML <intercept-url> drops servlet-path when building path matchers
Spring Security 7.0.0–7.0.4 silently discards the servlet-path attribute in XML authorization rules, enabling unauthenticated access to protected endpoints.
Cantina 9 min read
-
Vulnerability Research
Why current evidence matters in incident response
Current evidence enables faster incident response by making unusual execution paths queryable rather than requiring manual reconstruction.
Cantina 4 min read
-
Vulnerability Research
When the security tool becomes the attack path
Analysis of how the Checkmarx supply-chain incident demonstrates security tools can become vectors for attack when given privileged access to CI/CD environments and developer workflows.
Cantina 7 min read
-
Vulnerability Research
When a git helper inside automation becomes a remote-execution path
A vulnerability in simple-git demonstrates how helper libraries in trusted automation workflows can become serious security exposures.
Cantina 6 min read
-
Company News
One security OS, now plugged into EDR and MDM: Cantina partners with CrowdStrike and Swif
Cantina integrates endpoint detection and device management capabilities through partnerships with CrowdStrike and Swif.
Cantina 3 min read
-
Guides
Healthtech AI security: what to control before you ship
Framework for managing AI security risks in healthtech before product launch, covering PHI handling, code review, vendor management, and regulatory compliance.
Cantina 7 min read
-
Vulnerability Research
Your security stack doubled. coverage stayed flat
Fragmented security stacks grow in tools but fail to improve coverage due to manual cross-system context assembly.
Cantina 4 min read
-
Vulnerability Research
Your annual SOC 2 audit proves nothing about your security today
Annual SOC 2 audits provide historical snapshots but fail to address current security postures that regulators and buyers increasingly demand.
Cantina 4 min read
-
Vulnerability Research
13 Years of RCE in Apache ActiveMQ: what CVE-2026-34197 reveals about your blind spots
A critical RCE vulnerability in Apache ActiveMQ's management plane exposes how security gaps persist across organizational silos.
Cantina 4 min read
-
Vulnerability Research
Bitwarden's npm incident was a publish-path compromise
Analysis of how a malicious Bitwarden CLI package bypassed source controls through npm registry compromise.
Cantina 6 min read
-
Vulnerability Research
Can your security platform show who the agent was acting on behalf of?
A guide to evaluating agent accountability by examining whether security platforms can trace actions back to initiating users and their delegated authority.
Cantina 7 min read
-
Vulnerability Research
Healthtech integration risk: where PHI exposure becomes operational risk
Healthcare systems face substantial security exposure at data handoff points between EHRs, APIs, payers, and connected platforms.
Cantina 7 min read
-
Vulnerability Research
When checkpoints turn into code execution: how Apex would unwind LangGraph's deserialization trap
Analysis of CVE-2025-64439 in LangGraph showing how unsafe deserialization combined with serializer fallback logic creates remote code execution vulnerabilities.
Cantina 6 min read
-
Vulnerability Research
How production context changes which security alerts matter first
Production deployment context helps security teams prioritize alerts based on actual operational risk rather than severity alone.
Cantina 4 min read
-
Vulnerability Research
How auto-approve changes the control boundary for coding agents
Auto-approval settings for coding agents shift execution control boundaries and require careful policy definition before rollout.
Cantina 3 min read
-
Guides
How to catch secret leaks in AI coding workflows before commit
GitHub's MCP server push protection now blocks secrets within agent sessions rather than at repository commit.
Cantina 3 min read
-
Vulnerability Research
Cantina threat advisory: how Anthropic's MCP stdio model turns configuration into code execution
Analysis of how untrusted input reaching MCP stdio configuration can enable local code execution attacks.
Cantina 6 min read
-
Vulnerability Research
Vercel's April 2026 incident was an oAuth-to-control-plane breach
Analysis of how Vercel's April 2026 security breach leveraged compromised third-party OAuth access through Google Workspace to reach deployment control-plane configuration.
Cantina 11 min read
-
Vulnerability Research
Healthtech security in 2026: continuity Is the new standard
Security incidents in healthtech are now judged like operational failures, making continuity protection equally important as data protection.
Cantina 8 min read
-
Vulnerability Research
Apex finds 44-year-old bugs in OpenSSH
OpenSSH 10.3p1 fixes two vulnerabilities rooted in decades-old trust assumptions from Berkeley rcp.
Cantina 6 min read
-
Guides
AI agent governance readiness guide
Framework for assessing AI agent governance readiness across inventory, permissions, approvals, logging, and incident response.
Cantina 6 min read
-
Guides
How to prevent vibe coding vulnerabilities in software development
Automated code analysis helps teams counter "vibe coding" risks from AI-assisted development by catching security flaws early and reducing alert fatigue.
Cantina 3 min read
-
Guides
AI act transparency starts August 2, 2026: what AI-powered SaaS teams need to change now
SaaS teams must prepare now for EU AI Act transparency obligations taking effect August 2, 2026.
Cantina 5 min read
-
Guides
MCP security checklist: control what your AI agents can access
Secure MCP servers with 12 practical checks for permissions, OAuth, prompt injection, sandboxing and logging, plus tests to verify your controls.
Cantina 8 min read
-
Vulnerability Research
How one compromised GitHub action leaked thousands of cloud secrets
Analysis of the tj-actions/changed-files compromise that exposed secrets through mutable GitHub Action tags in 23,000+ repositories.
Cantina 4 min read
-
Product Updates
Cantina: unified security and compliance for AI-powered SaaS
Manual compliance evidence collection creates operational friction; continuous automated compliance reduces audit timelines and engineering overhead.
Cantina 3 min read
-
Vulnerability Research
SpEL injection in Spring AI explained
Spring AI's SimpleVectorStore is vulnerable to remote code execution through SpEL injection in filter expressions when keys are not properly sanitized.
Cantina 5 min read
-
Vulnerability Research
The hidden tax of tool sprawl: why your 50-tool security stack is failing you
Enterprises managing 45-83 security tools face fragmentation, alert fatigue, and financial losses that agentic AI platforms can resolve.
Cantina 11 min read
-
Vulnerability Research
Catching a 15-year-old dependency bug before attackers did
An AI-driven security tool discovered a 15-year-old buffer overflow vulnerability in XZ Utils before attackers could exploit it.
Cantina 3 min read
-
Vulnerability Research
Axios NPM supply chain attack: what happened
A threat actor compromised an axios maintainer account and published malicious npm releases that delivered cross-platform malware to approximately 3% of affected installations.
Cantina 3 min read
-
Vulnerability Research
Defending against FortiClient EMS pre-auth SQLi with agentic security
How agentic security platforms autonomously defend against CVE-2026-21643, a critical pre-authentication SQL injection vulnerability in FortiClient EMS.
Cantina 6 min read
-
Vulnerability Research
How Apex found a silent privilege escalation in Anthropic's Claude Code
A high-severity vulnerability in Claude Code allowed privilege escalation through workspace trust dialog bypass.
Cantina 3 min read
-
Vulnerability Research
How Apex found a high-severity bug in Spring AI
A high-severity Cypher injection vulnerability in Spring AI's Neo4j component was identified and autonomously remediated by Cantina's Apex security tool.
Cantina 3 min read
-
Vulnerability Research
How Apex found a critical RCE bug in Spring AI
Technical analysis of CVE-2026-22738, a critical Spring AI vulnerability enabling remote code execution through unvalidated SpEL injection in vector store filters.
Cantina 3 min read
-
Vulnerability Research
LiteLLM turned a package install into an intrusion path
A supply chain compromise in LiteLLM versions 1.82.7 and 1.82.8 delivered malware through PyPI that executed at Python startup via .pth files.
Cantina 6 min read
-
Guides
AI security: best solutions for cyber teams 2026 guide
A comprehensive guide addressing AI-powered defense tools and AI system security for SOC analysts, security engineers, and CISOs navigating 2026's threat landscape.
Cantina 18 min read
-
Vulnerability Research
The SOC analyst role Is broken. here's how agentic AI fixes it
Agentic AI transforms security operations by automating alert triage and investigation, freeing SOC analysts to focus on strategic work.
Cantina 6 min read
-
Guides
Securing AI agents: 5 rules to stop autonomous takeovers
Five AI agent security engineering rules for permissions, prompt injection, execution controls, monitoring, and incident response.
Cantina 6 min read
No articles match your search
Try another keyword or choose a different category.
Page 3 of 7