nx_known_issue_store.nx

nx_known_issue_store.nx -- LIB: the team's window onto the KNOWN-ISSUE catalogue in the SOVEREIGN INFORMATION STORE (knowledge/store/ki-*, via nx_seg_store). The pure-Nishi home for "have we seen this before?" -- NO TSV, NO SQL. The machine-matchable twin of the human-narrative landmines / workaround_audit registries: those are prose for a reader; THIS is bytes an organ matches a live failure against, so the Engineer/Doctor can RECALL a remedy instead of giving up. This is the structural twin of the "no-hang" error-emit guarantee (nx_netscope_dns recv-timed keystone): there, every error PATH must EMIT a verdict within a deadline (never hang). Here, every SEEN issue must be RECALLED + ROUTED to its remedy (never a silent "can't identify/fix this"). The _recall_or_red_gate locks it the way T2 locks no-hang. Record schema (all keys under prefix knowledge/store/ki-), TAB-separated value fields: ki:ids -> index: TAB-separated list of catalogue ids (the recall walk order) ki:<id> -> id <tab> class <tab> status <tab> signature <tab> remedy class = DIALECT|ENV|TOOLCHAIN|MATH (mirrors landmines scope) status = AUTO|ADAPT|DISCIPLINE|OPEN (the RESPONSE class) signature = exact substring to match in an Engineer diagnostic / crash artifact, or "-" when the issue has no crisp text signature (recall-by-id: a documented adaptation/discipline, not auto) remedy = organ:func for AUTO (e.g. nx_doctor_fix:doc_heal_token), else a response token (durable-runner-rearm, mask-after-shift, ...) ki:provenance -> authorship breadcrumb (staging-vs-verified is machine-visible in the store) license_tier: ORIGINAL