AI-generated video
Vibe Coding, code optimisation
and quality gates.
AI now writes large parts of the code — faster than teams can review it. This series covers what comes next: how AI-generated code can be measured, checked and secured, which metrics actually hold up, and at which point a gate is worth more than a review.
Claude Code Security and code review: how AI code is checked before it goes live
What Anthropic’s Claude Security and /security-review check, what they do not, and how you can automatically review every Claude Code diff and stop dangerous commands.
Code Guardian 17.17: Done Now Means: Goal Measured
What does Code Guardian 17.17 bring? A core probe blocks completion, the agent pauses after every adjustment, and small changes need less review. One gap remains open.
Code Guardian 17.16: Done only with proof, and a guard gap
When does a task count as done? From 17.15 on, only with proof. This overview of 17.14 to 17.16 also covers what the guards now catch and which gap remains.
Code Guardian 17.13: Done Means Proven
The agent reports “tests green,” yet the test run never happened. Code Guardian 17.13 no longer closes such a task, and shows how you can tell a proven result.
Claude Code Skills: Create, Install and Use Them Safely
What are Claude Code skills, where do they live, how do you install them, and when does a skill beat CLAUDE.md, a subagent or MCP? With a SKILL.md example and a security check.
Code Guardian 17.9: What Changes for You
Code Guardian 17.8 and 17.9 read commands that the gates previously missed: scripts behind redirections, more shell wrappers, remote ssh commands. Plus a one-time token check and a decision log for the council.
Code Guardian 17.7: What Changes for You
Since 17.6 there is the Leitstand for all Claude Code sessions, plus roughly 80% less of Code Guardian's own text per follow-up prompt. 17.4, 17.6 and 17.7 also close spellings that let commands slip past the guards. Here is what you should do.
Code Guardian 17.3: What Changes for You
Code Guardian 17.3 interrupts you less often. Twelve former blocks are now your own responsibility. This article lists which ones, how to regenerate approval reports, and what to check before the update.
Code Guardian 17.2: What Changes for You
Code Guardian 17.2 closes several ways around the safeguards, removes false alarms and makes check time measurable on Windows. Test coverage now counts more strictly, and the comparison measurement shows that a run with no difference is not proof that Code Guardian works.
Who Is Liable for AI-Generated Code? What Changes in 2026
As of December 2026, software is explicitly a product under product liability law. What this means for developers, agencies, and clients.
Definition of Done for AI Code: Evidence Instead of Promises
When code is written in minutes, the bottleneck shifts to the question of when something is actually done. A Definition of Done built for that pace.
Code Audit: A Checklist for Reviewing Someone Else's Software
Before a handover, before a purchase, after an incident: what a code audit checks, in what order, and how to spot a superficial one.
All images in this overview are AI-generated illustrations.AI transparency
Never miss an insight
Subscribe to our newsletter and get the latest articles straight to your inbox.
No spam. Unsubscribe anytime.
Ready for your AI competence certificate?
Get your AI certificate – flexible, online, with a documented certificate of participation for building AI literacy (Art. 4 EU AI Act).