nx_research_integrate.nx

nx_research_integrate.nx -- the INTEGRATED research/benchmark loop (operator: "integrate the fetch systems -- what we HAVE vs what we NEED vs more research, iterating toward s-class exceed, grounded in real tools like Unsloth (renamed from nx_research_ledger to un-shadow the distinct RESEARCH_MEMORY library runtime/nx_research_ledger.nx -- this is a research STATUS REPORT, that is an append-only findings LEDGER; two different capabilities, one name). that test speed + capabilities"). For each domain it joins: HAVE (a gate status-log exists?), RESEARCHED (counts the REAL fetched/*.raw corpus by prefix via getdents64 -- this is the fetch system integrated), the BENCH-TOOL (the incumbents' own yardstick to measure exceed by), and the NEED. Then it prints the prioritized NEXT actions. T1 the corpus is really read (sum of researched > 30). T2 >=6 domains have research. T3 the "tools that test speed+capabilities" ask is met: Unsloth (eff_) AND a bench-suite (bt_) are grounded. expect_exit: 0 Sovereign: nx_syscalls (getdents64 dir-read pattern from nx_claude_bridge).