nx_stepsolve_lib.nx

nx_stepsolve_lib.nx -- ES26 (engineshift al_steps): THE ONE RULER for STEPS-TO-SOLVE PER COMMON TASK. The operator's 2026-09-06 order: if a task can be solved in ONE step within logic it should be, and when it cannot the estate explains why two, why three, and so on, from the most common tasks to the least, mining and evaluating this efficacy continuously. The MEASURED side is the action journal's recurring tool:verb n-grams (nx_actlog's indexed miner, exact support, adjacent only): a recurring bigram is a 2-step ritual, a trigram a 3-step one, and its support is how common it is. The DECLARED side is knowledge/stepsolve.conf, DATA rows steps|<signature>|<min_steps>|<reason>|<one_step_verb> verbclass|<verb>|<reason> where signature = tool:verb>tool:verb[>tool:verb] exactly as the journal spells it, min_steps is the LOGICAL minimum, reason is why it cannot be fewer (a true dependency: a promote must see the sha only a finished build produces) and one_step_verb names the composed verb on the CANONICAL organ that does it in one call (nx_organ_ship for the build-stage-judge-promote ritual). A verbclass row is a RULE: every signature whose EVERY step carries that verb is explained by the rule's reason -- the clock plane and the job lane write <elf>:run for each organ fork, and a fork is one whole organ run and therefore one step by construction, so two adjacent scheduled forks are two tasks and not a ritual. A steps row for a specific signature WINS over a rule, so a fork ritual that does have a composed verb stays declarable. This lib joins the two and classifies every recurring signature: ONE-STEP-AVAILABLE min_steps 1 and a composed verb is named: the ritual is an ADOPTION gap, its support should fall DEPENDENCY min_steps equals the observed steps: the count is explained, the reason travels with it REDUCIBLE 1 < min_steps < observed: fewer steps are possible by declared logic BY-RULE no steps row, every step's verb matches a verbclass row: explained by that rule's reason UNDECLARED no row and no rule: NEVER assumed one-step; it is the worklist, most common first The published figure is a PARTITION over signature support (sums to the total, printed); per-rule support is tallied beside it so a rule can never hide what it absorbed; the gauge stamp is the EXPLAINED permil (declared classes over all), written in the CE9 heartbeat shape so gauge_pass renders it and a stale one reads STALE never zero. A row that cannot be parsed is counted as malformed, never dropped in silence. license_tier: ORIGINAL No hw writes.