Broken-Site Diagnostic Lab

You will diagnose four bounded incidents.

Use a disposable copy/branch. Do not corrupt the portfolio source you need.

Incident A: build stops

Introduce one known YAML or Liquid syntax failure.

You are not told which one in advance if your instructor/course lab provides the broken files.

Goal:

Identify the failing stage from terminal evidence before opening random files.

Incident B: build succeeds, page missing

Create a permalink/content-model/output configuration problem.

Goal:

Determine why expected destination output is absent.

Incident C: page exists, link 404s

Create a stale/wrong generated link.

Goal:

Compare generated href with actual destination/public path.

Incident D: development works, production model breaks

Use a baseurl or environment-dependent difference.

Goal:

Show which assumption changed between environments.

Required incident record

For every incident:

Symptom:
Build status:
First decisive evidence:
Failed layer:
Root cause:
Smallest repair:
Verification:

Scoring principle

A fast lucky fix with no diagnosis is weaker evidence than a slower, bounded diagnosis you can reproduce.

Checkpoint

Choose the incident that initially fooled you most. Explain what evidence corrected your first assumption.