OpenSpec Rejected Proposals: A Decision Memory Convention

An agent that proposed and shipped "move persistence into a shared library" six months ago will happily propose it again next quarter unless something durable tells it the idea was already investigated and rejected -- and OpenSpec has no built-in state for that today. /opsx:archive is built for one outcome: a change that shipped. It syncs the delta specs into openspec/specs/ and moves the folder to openspec/changes/archive/YYYY-MM-DD-<name>/ as a record of what changed and why. There is no /opsx:reject or /opsx:abandon counterpart, and nothing in the archive format tells a future proposal "this exact idea was investigated and turned down." That gap matters most in exactly the codebases where OpenSpec is otherwise a good fit: brownfield systems with a small number of contributors and agents that periodically re-explore the same architectural questions -- merge these two services, share this persistence layer, replace this HTTP boundary with a direct import. This is not a hypothe