Classic Unix command-line utility for searching plain text using pattern matching, commonly used for quick file and log searches.
One developer's 'could not tell' exit code stopped a false green on a hold, then exposed four new ways a check can mislead. Each looks like a clean result, so agent evals that adopt the third value need controls that force every outcome.
Reality
An engineer's three-day tally of his own tooling turned up about twenty checks that reported success while being structurally unable to report failure. The fixes mostly change what an exit code is allowed to mean.
Reality
A dev.to post logs nine defects found in one day's own verification tooling on 23 September 2026. The cheapest fix in it is one assert line that fires when a count of unique keys exceeds the rows it came from.
Reality
Pengcheng Xu's AgentConnect comparison finds semantic navigation beats grep only where text search is noisy; on clean repositories it added 16 to 19 percent to token cost for little or no accuracy gain.
Reality
A dev.to post argues the layers arrived before the problems. Its strongest evidence, grep beating an embedding pipeline, comes with no name and no numbers, which is the part worth testing yourself.
Reality