CLAUDE CODE AGENTS: FUNDAMENTALS - Chapter 10, Exercise 3
Why "Not Implemented Anywhere" Deserves Scrutiny
====================================================================================

QUESTION:
A research agent confidently reports that a particular feature "is not
implemented anywhere in the codebase." Using this chapter's own warning
box, explain why this claim deserves some scrutiny before being accepted at
face value.

SOLUTION / EXPLANATION:

This chapter's warning box is explicit that a confident answer from a
research agent is only as good as what its actual search covered - and a
negative claim like "not implemented anywhere" is, in a specific sense, an
even stronger and riskier claim than reporting where something WAS found.
Reporting a found location just requires one correct match; confidently
reporting that nothing exists anywhere requires the search to have
genuinely covered every place the feature COULD reasonably live, under
every naming convention it might reasonably use - a much higher bar to
actually meet.

If the codebase uses non-obvious naming for this particular feature (an
unusual internal term, a name inherited from an earlier version of the
project, or logic embedded inside a more general-purpose function rather
than a clearly-named dedicated one), a search built around the obvious,
expected terms could miss it entirely - and the agent would have no way to
know it missed anything, since from its own point of view, it searched
everywhere it thought to look and found nothing. The report "not
implemented anywhere" sounds exactly as confident whether that's actually
true, or whether the search simply didn't use the right terms to find it.

Given this, the reasonable response - matching the "trust but verify"
instinct this chapter borrows from Chapter 3 - is to ask what search terms
and locations the agent actually tried, and consider whether a differently
worded or broader follow-up search (perhaps explicitly requesting a
"thorough" breadth, per this chapter's own guidance) might be worth running
before fully accepting that the feature genuinely doesn't exist anywhere in
the codebase.

--------------------------------------------------------------------------
WHY THIS WORKS AS AN ANSWER:
It explains specifically why a negative claim ("not found anywhere") is
harder to substantiate than a positive one, and why it can sound equally
confident whether true or simply the result of an incomplete search,
connecting this directly to the chapter's own "trust but verify" guidance
for confident-sounding research results.