NewGitHub Package Firewall: Install-Time Security for AI Agents and Pipelines

Security,decoded.

Field notes on AI security, cloud architecture, Zero Trust and the vulnerabilities that matter. Written by William OGOU for builders and defenders.

Read the latest Browse 193 articles
Articles
193
Topics
809
Categories
10

Popular topics

TopicsBrowse by topic

All 809 topics

Latest

Fresh from the lab

The newest field notes on AI security, cloud and the vulnerabilities that matter.

View all

Topic atlas

Find your domain

10 categories, one map. Each bar shows a category's share of all 193 articles.

All topics

AI Security

Securing LLMs, agents and MCP: from prompt injection to AI-era supply chains.

65articles

34%

Cybersecurity

Threats, trends and strategy for security leaders and practitioners.

37articles

19%

Cloud Security

Hardening GCP, Azure and AWS: identity, data and network guardrails.

26articles

13%

Vulnerabilities

Critical flaws, active exploitation and the playbooks to respond.

23articles

12%

Zero Trust

Never trust, always verify: identity-first architectures, SASE and ZTNA.

7articles

4%

Network Security

Firewalls, edge and the network layer of defence in depth.

1article

<1%

Every topic

Browse the full tag index, from AI agents to zero-days.

809tags

Archive

From the archive

8 more recent reads, newest first. The full library has 193 articles.

  1. Cloudflare CA: Free Post-Quantum Certificates with MTCs in 2027

    Cloudflare plans free MTC certificates from 2027. See its new CA roadmap and current PQC services: Automatic Key Exchange, ML-DSA origin authentication, Tunnel, and certificate monitoring.

    Cybersecurity

  2. When Anthropic Talks About Its GLM Rival: The Cyber AI Line Has Moved

    Anthropic says rival open-weight model GLM-5.3 can build end-to-end exploits and has safeguards that are easy to bypass. What the tests show and what defenders should do.

    AI Security

  3. Mini Shai-Hulud Returns: Poisoned GitHub Actions Re-Enabled

    actions-cool/issues-helper and maintain-one-comment were re-enabled Sept 16 with malicious tags intact, re-exposing ~15K repos. Learn SHA-pinning fix.

    Supply Chain

  4. Is My Website Quantum-Safe? How to Check PQC in 5 Minutes

    Is your domain PQC compliant? Learn 3 quick ways to check post-quantum TLS (X25519MLKEM768) on any website and fix gaps before Q-Day.

    Cybersecurity

  5. Anthropic Fourth Breach: Claude Opus 4.6 Hacks Real Systems

    Anthropic disclosed a fourth incident where Claude Opus 4.6 breached real third-party systems. Learn how biased reasoning and recklessness drive AI escapes and how to harden evals.

    AI Security

  6. Cloudflare Post-Quantum TLS: Automatic Key Exchange Explained

    Cloudflare's Automatic Key Exchange learns each origin's key agreement preference and leads with post-quantum X25519MLKEM768 where supported. Learn how HelloRetryRequests fell from 52% to 3.7%, what to configure, and how to get your origins quantum-ready.

    Cybersecurity

  7. Google Mantis: Open-Source AI Harness to Find and Fix Bugs

    Google open-sourced Mantis, an AI agent harness that finds, reproduces, and patches vulnerabilities. Learn how the pipeline works, its stages, and how to get started.

    AI Security

  8. OWASP OASIS Is Official: AI-Powered Vulnerability Fixes at Scale

    OWASP OASIS is now an official OWASP community project. Learn how its AI-powered fix automation and human validation model aims to remediate open-source vulnerabilities.

    AI Security
Explore all 193 articles
Portrait of William OGOU

Meet the author

William OGOU

Google Cloud & AI Security Architect | CISSP, OSCP

Cybersecurity specialist and cloud architect building privacy-first AI tools. Here he writes about AI and LLM security, DevSecOps, cloud security on Google Cloud and security architecture.

  • AI Security
  • LLM Security
  • DevSecOps
  • Cloud Security
  • Google Cloud
  • Security Architecture
LinkedIn RSS All articles