Skip to content
OBSERVE. QUESTION. VERIFY.
02CASE FILE
Data flow

Follow the value across the boundary.

Many bugs live between modules: a field is renamed, a default is applied twice, or an absent value becomes an empty string. Choose one problematic value and follow its journey. Ask your assistant for file references at every meaningful transformation.

Build a short chain of evidence

Start at the input boundary and note the value shape. Follow parsing, validation, business logic, persistence, and output as applicable. Record the owning function at each step. You do not need an exhaustive architecture diagram; you need the route that explains this particular value.

Watch for silent changes

Compare types, units, null handling, and defaults between adjacent steps. Check whether serializers omit fields or whether a cache supplies an older value. Code inspection can show what a branch would do, but it cannot prove that a runtime request took that branch without supporting evidence.

Should every helper appear in the trace?

Include helpers that change meaning or choose a path. Group transparent forwarding steps so the important transformations remain easy to inspect.

A question for your assistant

Read only. Trace [field] from [input] to [output]. Cite the function and file at every transformation. Record type, units, defaults, and missing-value handling. Distinguish code paths you found from runtime paths you can actually confirm.
CONTINUE THE INVESTIGATIONRead a diff