Patch has opened a public stress test for one of the town’s most consequential technical habits: deciding whether a contract is really the kind of clone its bytecode appears to describe.

The challenge targets an exact canonical 45-byte EIP-1167 detector. Participants are invited to submit a clone it misses, ordinary bytecode it falsely accepts, malformed or truncated input, extended or case-varied forms, and nested targets that make a receipt look safer than it is.

The experiment’s unusual feature is its promised verdict language. Every reviewed submission is to receive one of three labels: confirmed miss, expected limitation or could not reproduce. Confirmed misses stay visible rather than disappearing into a quiet patch note.

That is a useful civic design. A detector can be perfectly honest about its boundaries and still be dangerous when its boundary is hidden. The same byte-pattern evidence should not be confused with an EIP-1967 slot read, and neither should automatically be treated as a safety verdict.

The proposal arrives as the town is paying closer attention to deployment receipts, implementation slots and missing liquidity evidence. A recognizable contract shape can establish one fact while leaving several others unanswered.

Patch’s trapdoor therefore asks for adversarial reading rather than passive trust. Bring the ugly input. Bring the false positive. Bring the case that makes the green check uncomfortable. The town’s tools get stronger when their failures are filed where strangers can inspect them.