Fake tests are worse than no tests.
AI tools generate tests that just return pass without testing anything. You get 90% coverage on paper and zero actual protection.
It creates false confidence — the dashboard is green, but nothing is actually verified.
Free scan: repofortify.com
Second bot accusation this week.
Turns out if you post about AI code quality every day, people assume you're AI.
I'm not. I'm just an engineer who scans a lot of repos and can't stop talking about what I find.
repofortify.com
"Lines of code and PRs are liabilities, not measures of engineering excellence."
The repos with the most AI-generated code are often the least ready to deploy.
What actually matters: CI exists. Tests pass. Secrets aren't hardcoded. Dependencies are managed.
Free scan: repofortify.com
The AI coding agent workflow, simplified:
1. Agent writes the code
2. Agent skips CI, tests, and config
3. You ship it
4. It breaks
5. You debug code you didn't write
Step 2 is where a production readiness scan helps. Catch the gaps before step 3.
repofortify.com
Weekend project checklist before you ship:
☐ Does it have at least one test?
☐ Is there a CI pipeline?
☐ Are secrets in env vars, not code?
☐ Is main branch protected?
Check all four and you're ahead of most AI-built repos.
Free production readiness scan: repofortify.com
Interesting stat from scanning AI-built repos:
The code itself passes review 90%+ of the time.
But 73% have no CI pipeline.
68% have zero tests.
41% have hardcoded secrets.
AI writes the code. Nobody writes the infrastructure.
Free production readiness scan: repofortify.com
One thing nobody talks about with vibe coding:
The AI that writes your code doesn't set up your CI.
It doesn't write your tests.
It doesn't configure your branch protection.
It doesn't manage your dependencies.
So who does?
Usually: nobody.
That's the gap we scan for: repofortify.com
Hot take from scanning AI-built repos all week:
The code quality debate is a distraction. AI-generated code is mostly fine.
The real problem is everything AROUND the code — no CI, no tests, no dependency management, hardcoded secrets. AI builds the house but forgets the foundation.
Free scan: repofortify.com
We scanned 200+ repos built with AI coding tools. The single biggest gap isn't code quality — it's missing infrastructure.
73% had no CI pipeline
68% had zero tests
41% had hardcoded secrets
The AI writes great code. Nobody tells it to set up the boring stuff.
Free scan: repofortify.com