Report absence of evidence
"I found nothing" hides the gap between "it is not there" and "I did not look properly". This skill makes the scope and the sensitivity of a null result explicit so no one mistakes it for proof.
Procedure
Record the exact command that found nothing:
grep -rn "api_key" src/ --include=*.py.Record the scope implied: which directories, file types, hosts, branches, and time range were actually covered.
Capture the negative result verbatim, including the exit code.
grepandrgreturn 1 on no match,findexits 0 with empty output — the code is part of the evidence.Test the search's sensitivity. Would it have matched if the thing were present? A case-sensitive
grepmissesAPI_KEY, so re-run with variants:grep -rin "api[_-]\?key" src/.Run a positive control where possible: search for a string you know exists, to prove the tool and path work.
Report as "no match in using ", never as "there is no X".
Name the parts of the space you did not search — other branches, the database, the vendor's closed source — so the limit of the null is explicit.
Pitfalls
- Searching the wrong branch or a stale checkout is the most common false negative; confirm with
git rev-parse --abbrev-ref HEAD. - A tool's include filter silently narrows scope;
--include=*.pyskips config files, so say so. - Absence in logs may mean logging is off, not that the event never happened.
- Reporting a null as a clean success skips the case where the check itself is broken.
- A regex typo produces a confident empty result; the positive control is what catches it.
Verification
grep -rn "api_key" src/ --include=*.py; echo "exit=$?" # exit=1 is the no-match signal
Report scope, command, and exit code; say "not found in X", never "does not exist".