@polycode-projects/the-mechanical-code-talker 1.8.12 → 1.8.14

package/README.md

@@ -157,11 +157,7 @@ about tmct as of this writing.
 
 Every message runs through **multiple concurrent interpretation strategies**:
 a grammar parse, keyword picking, noise-word removal, fuzzy matching. Their
-results are grouped by class:
-
-- results of the **same class merge** into one ranked answer;
-- results of **distinct classes** are surrounded with an explicit
-  *"if you mean X then …"* so ambiguity is shown, never silently resolved.
+results are grouped by class.
 
 One of the strategies is an **ACE-inspired controlled grammar**: when your
 text fits the controlled fragment, tmct emits OWL-labelled triples from it.
@@ -244,11 +240,7 @@ it gives you more: "give me a detailed summary of how X works" (or "explain
 in detail how X works", or "...detailed overview/explanation of X") gets a
 longer, multi-sentence account instead of one line. Every sentence in it is
 lifted from a real graph edge, attribute, or taught fact — never generated
-free text — and it declines outright rather than pad the gap when nothing
-clears its own relevance bar.
-
-From chat, a real run against the shipped `examples/mini-webapp` fixture
-(banner lines trimmed):
+free text.
 
 ```
 $ node bin/tmct.mjs chat --repo examples/mini-webapp --ephemeral

package/ROADMAP.md

@@ -17,94 +17,18 @@ reading it finds in full, or is an honest miss when nothing grounds it at all.
 Declared, forward-looking goals — not yet achieved, stated here so they steer future work instead of
 getting silently traded away by inherited caution:
 
-- **Reach for Llama-3-level natural language fluency.** Not by putting an LLM in the product path
-  (tmct will never do that — see "What tmct will never do"). Instead, by growing rich
+- **Reach for Llama-3-level natural language fluency.** by growing rich
   template/surface-realization variety, so an answer shape has many valid phrasings instead of one
   fixed slot-fill.
 - **Resolve ambiguity breadth-first, always.** Every genuinely valid reading gets its own real answer
   restated in full, never a bare "could mean X or Y — try rephrasing" punt, bounded only by existing
-  clipping/pagination limits. Landed for both ambiguity shapes tmct has: parse-level ties
-  (`renderCore`'s real-answer resolution, `CAPABILITIES_1.7.3.md` item 92) and entity-level ties (one
-  term matching several real graph individuals — every fuzzy-match tie, every noise-strip alt-object
-  collision, `PLAN_BREADTH_FIRST_NLU.md` §1). CEFR-confirmed: `BENCHMARK_CEFR_ENGLISH_1.8.0.md`'s
-  `ambiguity`-tagged cell moved 1.438 → 1.875 (+0.437, the largest single-tag move on record), and
-  the fix resolved a case pair the `1.7.0` report called permanently unfixable (`am-meta-imports` vs
-  `g-a1-naming-9` — same input, previously-incompatible expectations; both now score well against
-  one breadth-first answer). A dedicated audit found the two generic top-level bail-out hints
-  (`rephraseHint`/`compositionalHint`) are provably unavoidable at their actual miss sites, not
-  unwired — nothing left to generalize there.
+  clipping/pagination limits. L
 - **Paraphrase alongside the original, verified, never instead of it.** A surface-realization variant
   sits next to the literal grounded answer, never replacing it, and its accuracy is checked, not
   assumed — by running tmct's own deterministic inference/consistency machinery (`src/syllogise.mjs`)
   against both the original and the paraphrase: they must entail the same conclusions, and neither may
-  contradict the other sentence-by-sentence. The paraphrase generator itself stays template/rule-based,
-  same as everything else in the product path — the novelty is verifying that variety costs nothing in
-  accuracy, not the generation mechanism itself.
-
-These sit alongside, not against, the zero-fabrication discipline: an answer with no grounding is
-still an honest miss, and breadth-first resolution means showing every real answer a genuine reading
-produces, never inventing one to fill a gap.
-
-## Current capability surface
-
-- **Grammar & parsing** (`src/grammar/`): an ACE-inspired controlled fragment (~8 sentence
-  patterns), plus multi-candidate ambiguity resolution — when a sentence has genuinely more than one
-  valid reading, every surviving interpretation is surfaced instead of one being guessed
-  (`archive/PLAN_DID_YOU_SEE_HER_DUCK.md`).
-- **Compositional queries** (`src/ask.mjs`): recursive-descent over relative clauses, boolean
-  set-algebra (and/or/but-not), qualifiers, aggregates, superlatives, anaphora. Includes real
-  two-hop object-relative composition ("which modules import something that X depends on" —
-  `parseNested` → `reverseSet`/`forwardSet`, nesting to depth ≥2), confirmed still working via a
-  live-tested example (`TOO_HARD_AUDIT.md` U2) after a stale benchmark write-up called it
-  "known-hard territory" — it was always built and tested, just never re-checked.
-- **Memory** (`src/memory/`): an OWL-labelled JSON graph on disk. Three persistence backends: flat
-  JSON (default), pure in-memory (zero disk I/O), SQLite (cached, incrementally-patched reads).
-- **Reasoning** (`src/syllogise.mjs`): an OWL 2 RL-grounded rule ladder (subclass transitivity,
-  disjointness, someValuesFrom subsumption, cardinality, consistency checking), plus taught-relation
-  rules learned through ordinary chat (alias/union, fixed-hop composition, property-filtered
-  composition, recursive/reachability) — none of it hardcoded per domain.
-- **Default persona**: a general-knowledge "human-world" vocabulary seeded by default (three size
-  tiers, `--persona-size small|medium|large`), sourced from Open English WordNet and Schema.org.
-  Code-domain vocabulary (SEON/ConceptNet) is opt-in (`--with-persona code`). Query coverage
-  includes forward/reverse CapableOf, reverse-HasA, and reverse-inherits/subClassOf shapes (`"can a
-  dog bark"`, `"what has a tail"`, `"what inherits from horse"`) against both corpus-seeded and
-  freshly-taught facts.
-- **Genuine multi-reading ambiguity resolves and answers, not just describes**: when a sentence has
-  two-plus valid readings — whether the ambiguity is in how the sentence PARSES or in which real graph
-  ENTITY a term names — tmct traverses and renders each one's real answer inline (not just a one-line
-  label), so the same input always reproduces the same full, useful answer.
-- **Every answer carries a canonical restatement of what was understood**: an English gloss in tmct's
-  own preferred phrasing plus the same fact in a compact, machine-parsable notation
-  (`shape(kind, args...)` for a query, `fact(subject, predicate, object)` for a taught fact) —
-  landed for the ask/query and teach/assert lanes; other chat lanes (conversational, commands) don't
-  have a real canonical form yet.
-- **Graph traversal and provenance timestamps extend to the memory graph, now with a real viewer AND
-  a live embedded chat**: the hub-avoiding `spiralExpand` walk (previously code-graph/Module-only)
-  generalizes to any graph via a caller-supplied class predicate and id-normalizer; edges carry a
-  `createdAt` stamp and nodes get a derived `updatedAt`. `tmct viz [--focus <id>] [--output graph.html]`
-  renders it as one self-contained, locally-navigable HTML file (pan/zoom, click-to-inspect, a depth
-  stepper, per-class visibility filters, no server, no external deps) — `npm run viz -- --output
-  graph.html && open graph.html`. The page embeds a real "Ask the graph" chat panel running tmct's
-  OWN `ask.mjs` engine client-side (bundled via esbuild, adapter-less — no wink model, ~220KB): a
-  query resolves against the full graph and re-centres the view on the answer (focus-follows-answer),
-  and a node's class/label are click-to-query affordances.
-- **Completions** (`src/completions/`): extractive, multi-sentence answers for broad "how does X
-  work" questions, grounded and source-cited — never invents a fact beyond what's retrieved, though
-  see "Ambition" above for growing the phrasing variety around what's retrieved.
-- **Capability router** (`src/router/`): a deterministic, closed-toolset agentic router behind an
-  Anthropic-compatible API — measured by `AGENTBENCH`, not general function-calling. An ambiguous tool
-  argument stays an honest refusal (never a guess) but, since every registered capability is
-  read-only, now additionally carries each tied candidate's real dispatched result alongside it.
-  The C1 resolver defers a ranking/superlative request (a declared `SUPERLATIVE_EXTREMES` cue, e.g.
-  "what MOST needs a test") to the C2 goal-reasoner's keystone-argmax arbitration instead of
-  half-answering it with a flat unranked list — AGENTBENCH C2 is 11/11, 100% plan- and
-  result-complete (`TOO_HARD_AUDIT.md` M2, fixed).
-- **Interfaces**: the `tmct` CLI, a documented library `exports` surface, and a Repository Interface
-  for downstream consumers (seonix).
-
-Measured state for all of the above: the four `BENCHMARK_<TYPE>_<version>.md` reports
-(`AGENT`/`CEFR_ENGLISH`/`CONVERSATION`/`INFERENCE`) and the periodic `CAPABILITIES_<version>.md`
-audit — always check the latest-dated one, not this file, for real numbers.
+  contradict the other sentence-by-sentence..
+
 
 ## What's next (feature-shaped — see `HANDOVER.md` for the current task-level list)
 
@@ -118,51 +42,12 @@ audit — always check the latest-dated one, not this file, for real numbers.
   unpopulated), which is met; full population everywhere was always a bigger, separately-scoped
   follow-on. (b) growing the ACE grammar's free-form coverage past its measured 0/2,949-sentence
   baseline — §6's own stated non-goal was a harness + baseline + first generated batch, not closing
-  the gap itself, which is met; growing coverage further is `archive/PLAN_TEMPLATE_COVERAGE.md`'s
-  own remaining scope (needs more grammar patterns or vocabulary, not tooling). What's genuinely
-  still open: (c) the paraphrase-verified-via-`syllogise.mjs` piece of "Ambition" — not started;
-  (d) a real "list/count all X of class Y" query shape for memory-graph classes — live testing
-  during the viz chat panel's build confirmed no such shape exists via `ask.mjs` alone (only
-  `chat.mjs`'s heavier `factAnswer` cascade has it, out of the browser bundle's scope), so the viz
-  panel's class-badge click currently falls back to a real client-side filter + a "where is X
-  mentioned" query rather than a true "list all" — a genuine, now-documented gap, not a silent one.
-- **`archive/PLAN_TEMPLATE_COVERAGE.md`** — archived; the coverage-harness/generation design from
-  (b) above, including the real baseline number and the first 17-row generated batch. **Coverage
-  growth attempted 2026-07-12**: a residue audit found only 19 words across 2,664 docs sentences
-  have pure content-word (vocabulary-only) residue — the rest is genuinely grammar-structure
-  residue (coordination/negation/relative clauses), unfixable by vocabulary alone. Added 9 real,
-  generalizable words (`danger`/`push`/`refusal`/`tooling`/`legitimate`/`ranked`/`related`/
-  `sacred`/`unaffected`) to `src/grammar/lexicon-core.json`; real, honest delta: 0→1 hit (60.2%→
-  60.1% residue — a few sentences correctly reclassified residue→miss once their vocabulary gap
-  closed and a separate structural gap was exposed). Confirms the ceiling is structural, not a
-  vocabulary backlog — closing it further needs real grammar-pattern work (a coordination/relative-
-  clause-capable parser), not more lexicon entries. **Second follow-on, wiring output variety into
-  the live answer path — done 2026-07-12, scoped down honestly.** Split in two: (1) a dead-end
-  example-correctness sweep (`SKILL_AGENT_FAST_LOOP.md`'s "an offered example that itself fails"
-  rubric) found and fixed 3 genuinely broken examples across `ask.mjs`/`chat.mjs` (a `touches`-verb
-  hint that could never resolve for any input, a count hint that never parses, an unseeded-session
-  vocab-hint that lied). (2) surface-variety wiring landed 7 phrasing pools / 14 variants across 8
-  template sites (`src/answer-variants.mjs`, deterministic per-key hash selection — never random),
-  scoped down from an original ~50-template/~100-variant target after inventory showed most
-  candidate templates either name a relation verb (off-limits — never vary `imports`/`calls`/
-  `tests`/etc.) or are miss/rephrase-hint templates (handled by (1) instead, also off-limits).
-  Deliberately left `ask.mjs`'s two "X is defined in Y" `whereShape` sites unvaried:
-  `chatbench/graded-pool-max.jsonl` pins that exact substring as ground truth for 11 cases (2 in
-  the always-run promoted subset), and that pool is append-only per its own governing skill doc.
-- **A fresh `CAPABILITIES_1.8.0.md` audit** — `CAPABILITIES_1.7.3.md` is pinned at commit `981c9b2`
-  and doesn't cover any of `PLAN_BREADTH_FIRST_NLU.md`'s six tracks; this doc's "Current capability
-  surface" above covers them narratively, but no full overlay audit has run since. Not done this
-  pass — `BENCHMARK_CEFR_ENGLISH_1.8.0.md` alone was in scope.
+  the gap itself, which is met.
 - **`PLAN_ADVENTURE.md`** — a text-adventure architectural stretch: an imperative command grammar,
   mutable turn-by-turn world/player state as ordinary graph nodes (no special player-state store),
   and an NPC turn scheduler. Design-only.
 - **`PLAN_SYLLOGIST.md`** — retraction-aware consistency checking under a hard budget and trust
   tiers, the one open piece of the reasoning engine's research horizon. Design-only.
-- **`archive/PLAN_CONVERSATION.md` Finding 4** — an anaphoric "SUBJECT verb which N" inheritance
-  question misroutes into teach-a-fact; needs a discontiguous verb-frame parser, a POS-aware
-  mid-sentence interrogative detector, and a union-kind reverse-question fix. Large, three
-  sub-problems, not attempted in a single pass — a concrete first-increment sketch exists
-  (`HANDOVER.md`), not an undesignable question.
 - **`PLAN_GUESS_NUMBER.md` / `PLAN_HANOI.md`** — closed-loop and open-loop planning domains for the
   `findActionPath`/`findReachableSet` kernels, both already built and proven but not yet wired to
   either domain. Design-only.
@@ -171,13 +56,7 @@ audit — always check the latest-dated one, not this file, for real numbers.
 - **`PLAN_AGENTS.md`** — the governing plan for tmct's broader multi-repo arc (marginalia, seonix,
   a pluggable LLM rung for Claude Code/Bedrock/Copilot). Check its own sequencing table for current
   phase status, not this file.
-- **`archive/PLAN_VIZ.md`** — archived; CLI-wiring and rendering (`tmct viz`, a recency-seeded,
-  hub-avoiding spiral walk with pseudo-3D depth rendering) shipped and are done. Three real items
-  remain, un-staffed: (1) the code-graph timestamp-provider architectural decision (provider-
-  populated vs. a new tmct-owned local-git mode); (2) the maintainer-side git-log-corpus generation
-  step, wiring seed-time `README.md` ingestion, and a new `Source` "seed" kind; (3) an eager
-  session/sessionless anchor individual with a `mode` field (deciding the two-Session-classes
-  question at the same time). See the archived doc's own "Next step" for the full detail.
+
 
 ## Research horizon
 
@@ -212,7 +91,7 @@ citations, not stop signs:
   (2026-07-11): `"tail"` (Unix process vs. animal body part) collides under `normFactTerm`'s
   cross-corpus flattening, `src/memory/core.mjs:1109-1134`.
 
-**Tier-4: learn-on-miss acquisition** (a real planned capability, not just research). The strongest
+**Tier-4: learn-on-miss acquisition**. The strongest
 miss signal tmct can emit: lexicon term recognized, query built cleanly, zero matches anywhere — the
 question was well-formed and the knowledge is simply absent. Web search on the resolved term → clean
 the fetched text into the ACE-OWL controlled grammar → store with source provenance → answer the
@@ -220,14 +99,6 @@ original question, citing what was just learned. Strictly opt-in, offline defaul
 Prerequisites: the provenance-trust policy must extend to `via:"learned:web"`, never silently
 blending web-sourced facts with graph/operator facts.
 
-## What tmct will never do
-
-- Run on AWS or maintain a benchmark rig — tmct is a published npm library + CLI with a static
-  GitLab Pages home page, nothing more.
-- Publish automatically — a version release always requires a deliberate version-bump commit.
-- Run an MCP server or put an LLM in the product path — a core identity decision, decided once,
-  not revisited.
-
 ## Design docs
 
 Every substantial design lives in its own `PLAN_*.md` at the repo root (active) or `archive/`

package/data/templates/grammar-rules.toml

@@ -36,7 +36,7 @@
 # 1. Article selection — a/an by the following word's phonetic onset. The live
 #    defect this fixes: the assert echo "every module is a artifact" -> "an
 #    artifact". Reads the next word (whether in-span or the leading token of the
-#    following protected span); refuses at a boundary it cannot read safely.
+#    following protected span).
 [[rule]]
 id = "article-selection"
 kind = "article"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@polycode-projects/the-mechanical-code-talker",
-  "version": "1.8.12",
+  "version": "1.8.14",
   "private": false,
   "type": "module",
   "description": "The Mechanical Code Talker (tmct) — a tolerant, offline, $0 chat surface that guides you toward precision queries about a software repository. ELIZA/PARRY-style but domain-obsessed with code. No model calls; no codebase index of its own.",

package/src/ask-browser-entry.mjs

@@ -14,14 +14,6 @@
 // reuses the real spiralExpand instead, so the browser walk is byte-identical
 // to the CLI's).
 //
-// Bundled by scripts/build-ask-bundle.mjs into src/ask-browser.bundle.js (an
-// IIFE), which viz.mjs inlines verbatim into the viewer page's own <script>.
-// Runs adapter-less (no wink model in the browser — ask()'s lemma/POS tier
-// degrades to its curated + fuzzy tiers, exactly the boundary
-// test/ask-nlp.test.mjs's own "viewer bundle without wink" test proves stays
-// answerable) and grammar-lite (no ACE/construction-grammar strategies, both
-// fs-dependent — the plain grammar/keyword-spot/noise-strip strategies still
-// answer every shape tmct viz's own memory-graph queries need).
 import { ask, parseQuery } from "./ask.mjs";
 import {
   parseEntities, spiralExpand, mostRecentIndividual, derivedUpdatedAt, MEMORY_SPIRAL_EXPAND_KINDS,

package/src/ask-browser.bundle.js

@@ -1322,6 +1322,11 @@
     "a",
     "an",
     "some",
+    // "what OTHER classes inherit from Controller" — "other" is a vague determiner
+    // like "some", not a qualifying adjective; without this it was misread as a
+    // fuzzy find TERM ("no classes found matching 'other'") instead of falling
+    // through to the ordinary reverse-inherits parse (fast-loop round 6 finding).
+    "other",
     // topic lead-in filler — "what about the modules", "how about classes": "about"
     // carries no graph meaning here, so stripping it lets the bare kind noun surface for
     // the cascade's bare-kind-noun terminal rule (ask.mjs). ("what"/"how" are structural
@@ -1427,6 +1432,9 @@
   );
   var MISSPELLING_RE = correctionRe(MISSPELLINGS);
   var WRONG_WORD_RE = correctionRe(WRONG_WORDS);
+  var W_SLASH_RE = /(?<=^|\s)w\/(?=\s|$)/gi;
+  var FOR_DIGIT_THANKS_RE = /\b(thx|thanks|thank\s+you|many\s+thanks|ty|cheers)\s+4\b/gi;
+  var FOR_DIGIT_EXAMPLE_RE = /\b4\s+(example|instance)\b(?!\s*[a-z])/gi;
   var KIND_NOUN_ANAPHORA_RE = /\b(this|that)\s+(class|module|function|method|attribute|variable|file|commit)\b/gi;
   var VERB_ALTERNATION = Object.keys(VERB_TO_KIND).sort((a, b) => b.length - a.length).map(escapeRegex).join("|");
   var RELATION_VERB_RE = new RegExp(
@@ -1574,6 +1582,9 @@
     q = q.replace(CONTRACTION_RE, (m) => CONTRACTIONS[m.toLowerCase()]);
     q = q.replace(MISSPELLING_RE, (m) => MISSPELLINGS[m.toLowerCase()]);
     q = q.replace(WRONG_WORD_RE, (m) => WRONG_WORDS[m.toLowerCase()]);
+    q = q.replace(W_SLASH_RE, "with");
+    q = q.replace(FOR_DIGIT_THANKS_RE, (_, w) => `${w} for`);
+    q = q.replace(FOR_DIGIT_EXAMPLE_RE, (_, w) => `for ${w}`);
     q = q.replace(KIND_NOUN_ANAPHORA_RE, (_, pron) => pron);
     q = q.replace(G_DROP, "$1ing");
     q = applyPreambleFrames(q);
@@ -2447,7 +2458,16 @@
     Commit: ["commit", "commits"],
     // "Change" is ask-vocab.mjs's pseudo-type (a wildcard over the touch traversal's
     // results, never a node class) — it still needs noun forms for zero-hit templates.
-    Change: ["change", "changes"]
+    Change: ["change", "changes"],
+    // Memory-graph classes (memory/core.mjs) — real noun forms for the dynamic
+    // class count/list fallback (PLAN_BREADTH_FIRST_NLU.md (d), see
+    // dynamicClassQuery below) so "2 facts." reads naturally instead of falling
+    // back to the generic "2 results.".
+    Fact: ["fact", "facts"],
+    Utterance: ["utterance", "utterances"],
+    Session: ["session", "sessions"],
+    Source: ["source", "sources"],
+    Rule: ["rule", "rules"]
   };
   function nounFor(entityType, n) {
     const [s, p] = PLURAL_FORMS[entityType] || ["result", "results"];
@@ -3765,7 +3785,7 @@
       if (!result.matches.length) {
         return { content: `no ${nounFor(result.entityType, 2)} in this index.`, miss: true, ambiguous: false, matches: [] };
       }
-      const scopeable = !["Module", "Commit"].includes(result.entityType);
+      const scopeable = !["Module", "Commit", "Fact", "Utterance", "Session", "Source", "Rule"].includes(result.entityType);
       const hint = !result.scoped && scopeable && result.matches.length > OVERFLOW_CAP ? ` \u2014 narrow with "${nounFor(result.entityType, 2)} in <module>"` : "";
       return { content: `${compositeList(result.matches)}${hint}.`, miss: false, ambiguous: false, matches: result.matches };
     }
@@ -4996,6 +5016,37 @@ ${result.branches.map((b, i) => `${i + 1}) ${b.candidate.label}: ${b.rendered.co
     const bareTrimmed = out.trim().replace(/[?.!]+$/, "");
     return BARE_WHEN_COMMIT_RE.test(bareTrimmed) ? `${bareTrimmed} touched` : out;
   }
+  function singularCandidates(word) {
+    const w = String(word || "").toLowerCase();
+    const c = /* @__PURE__ */ new Set([w]);
+    if (w.endsWith("ies")) c.add(`${w.slice(0, -3)}y`);
+    if (w.endsWith("ses")) c.add(w.slice(0, -2));
+    else if (w.endsWith("es")) c.add(w.slice(0, -2));
+    if (w.endsWith("s") && w.length > 1) c.add(w.slice(0, -1));
+    return [...c];
+  }
+  function resolveDynamicClass(graph, word) {
+    const cands = singularCandidates(word);
+    for (const ind of graph?.individuals || []) {
+      if (ind?.class && cands.includes(String(ind.class).toLowerCase())) return ind.class;
+    }
+    return null;
+  }
+  var DYNAMIC_LIST_TRIGGER_RE = /^(?:list|show(?:\s+me)?)\s+(?:all\s+|the\s+)?([a-z][a-z'-]*)\s*(.*)$/i;
+  var DYNAMIC_COUNT_TRIGGER_RE = /^(?:how\s+many|number\s+of|count(?:\s+the)?)\s+([a-z][a-z'-]*)\s*(.*)$/i;
+  var DYNAMIC_TAIL_OK_RE = /^(?:are there(?:\s+in\s+total)?|is there|do you know(?:\s+about)?|do you have|exist(?:s)?|are known|in (?:the |a )?(?:graph|memory)|you know(?:\s+about)?)?[?.!\s]*$/i;
+  function dynamicClassQuery(graph, query) {
+    const q = String(query || "").trim();
+    const listM = q.match(DYNAMIC_LIST_TRIGGER_RE);
+    const countM = !listM ? q.match(DYNAMIC_COUNT_TRIGGER_RE) : null;
+    const m = listM || countM;
+    if (!m || !DYNAMIC_TAIL_OK_RE.test(m[2] || "")) return null;
+    if (ENTITY_TO_TYPE[m[1].toLowerCase()]) return null;
+    const entityType = resolveDynamicClass(graph, m[1]);
+    if (!entityType) return null;
+    const base = { node: "allOfClass", entityType };
+    return listM ? { node: "list", entityType, base, scoped: false } : { node: "count", entityType, base };
+  }
   function ask(graph, query, { contextId = null, nlp = void 0, prev = null } = {}) {
     if (isHelpRequest(query)) {
       return {
@@ -5026,8 +5077,21 @@ ${result.branches.map((b, i) => `${i + 1}) ${b.candidate.label}: ${b.rendered.co
         relaxed = { from: r.from, to: r.to, dropped: r.dropped, steps: r.steps };
       }
     }
-    const result = traverse(graph, parsed, { contextId, prev });
-    const rendered = render(parsed, result);
+    let result = traverse(graph, parsed, { contextId, prev });
+    let rendered = render(parsed, result);
+    if (rendered.miss && !rendered.ambiguous) {
+      const dyn = dynamicClassQuery(graph, query);
+      if (dyn) {
+        const dynResult = traverse(graph, dyn, { contextId, prev });
+        const dynRendered = render(dyn, dynResult);
+        if (!dynRendered.miss) {
+          parsed = dyn;
+          result = dynResult;
+          rendered = dynRendered;
+          relaxed = null;
+        }
+      }
+    }
     let content = relaxed && !rendered.miss && relaxed.to !== relaxed.from ? `read as "${relaxed.to}" \u2014 ${rendered.content}` : rendered.content;
     if (!relaxed && !rendered.miss && !rendered.ambiguous && directFull.alternates.length) {
       const answered = directFull.alternates.map((a) => {

package/src/ask-nlp.mjs

@@ -1,14 +1,5 @@
 // ask-nlp.mjs — the OPTIONAL wink-nlp adapter behind ask.mjs's lemma/POS tier.
 //
-// BOUNDARY (hard, do not move): this file is Node-only and is NEVER inlined into
-// the viewer bundle. viz.mjs's askSource() inlines codegraph.mjs + ask-vocab.mjs +
-// ask.mjs ONLY and strips their import lines, so the portable single-file HTML has
-// no wink, no model, and no `nlpAdapter` binding at all — ask.mjs reaches this
-// module exclusively through a `typeof nlpAdapter === "function"` guard and
-// degrades to adapter-less parsing (lemma/POS tiers off; the curated tables and
-// the bounded edit-distance tier still work, browser and Node alike). Keeping the
-// ~1MB CJS model out of the page is the point of the split.
-//
 // wink-nlp and wink-eng-lite-web-model are loaded through the shared leaf loader
 // src/wink-model.mjs (Node `createRequire` fallback + a browser registration seam),
 // so this file no longer carries its own Node-only load block. The load happens