
Threat reports and
security research notes
Research Lab reviews publicly documented attacks and ransomware trends.
When we reproduce a technique, the records we publish contain no customer data.
Reports, advisories and guides

Four paths for MCP supply chain risk and what to check

Protecting backups and recovery paths from ransomware

Which decisions should stay with people in an AI-enabled SOC?

Mitigating the Langflow vulnerability CVE-2026-33017

Mitigations for three vulnerabilities in the Anthropic Git MCP server

What to retain when auditing AI security verdicts

Security checks for Korean-language input handling

What npm supply chain incidents reveal about installation risk

Designing customer data isolation for security AI

How to pilot an AI-enabled SOC, with a sample 90-day plan

AI security validation checks and evaluation criteria
No articles match. Try a different search term.
Get threat report updates by email
Open the subscription request in your email app and send it to support@genox.one. The draft includes your consent to receive updates. You can unsubscribe at any time.
Checks before deploying AI agents
These notes organize checks using public specifications and OWASP risk categories.
They contain no customer incidents or measurements.
Prompt injection
Distinguish direct instructions in chat from untrusted instructions in documents. Require human review before consequential actions.
MCP tool descriptions
Review names and descriptions before connecting tools. Verify a server's read-only claims independently.
Tool permissions
Limit unnecessary access and review approval rules for payments, deletions and email.
Packages and model files
Verify names, versions and hashes against the lockfile, and confirm the source of model files.
Reference guides
Threat groups targeting South Korea
Public sources describe five groups, including Lazarus and Kimsuky, and what to check in detection.
Korean regulatory guides
Read about ISMS-P, financial network separation, N2SF, the AI Basic Act and personal data security measures from a security operations perspective.
Security glossary
Practical examples explain 40 common terms, including red teaming, EPSS, MCP and network separation.