Evidence Without Precision Theater

Portfolio evidence is a promise. A source link says the work can be inspected. A deployment link says the work can be used. A test reference says the behavior has a guardrail.

But evidence can also become theater. The moment a portfolio card hard-codes a test count, an old deployment ID, or a stale commit, it starts borrowing credibility from a fact that is already decaying.

#Evidence is not a dashboard

The page should not pretend to be a live operations console unless it actually has a live data source. For a personal studio, evidence should answer simple questions:

• Where is the source?
• Where is the running surface?
• What behavior is covered?
• What changed because of this work?

That is enough for trust. More precision is not automatically more truth.

#Volatile facts need a home

Deployment IDs, test counts, and commit hashes should live in release traces, CI artifacts, or generated release evidence. They can still be shown, but they should not be hand-maintained inside MDX.

The rule is blunt: if a fact changes frequently, it needs a system. If it does not have a system, describe the stable coverage instead.

#The better card

A better evidence card says: “Playwright covers public routes, command flows, copy feedback, filters, and mobile overflow.” That is less numerically exciting than a count, but it is more durable.

Precision is only valuable when the maintenance loop is real.