@adia-ai/a2ui-mcp 0.4.7 → 0.4.8

package/CHANGELOG.md

@@ -11,6 +11,18 @@ zettel strategies.
 
 _No pending changes._
 
+## [0.4.8] - 2026-05-12
+
+### Changed — eval threshold rebaseline + smoke-script alignment (§87, v0.4.8)
+
+Companion to `@adia-ai/a2ui-compose@[Unreleased]` §87 zettel `ensureBooted()` race fix.
+
+- **`mcp/scripts/eval-diff.mjs`** — thresholds rebaselined to the honest post-§72 floor (~5%; was holding stale v0.4.6 baseline). Eval coverage gate now reflects the chunks-only retrieval shape after the patterns retirement.
+- **`mcp/scripts/smoke-engine-registry.mjs`** — retrieval-quality probes updated to match the post-§72 + §88 retrieval surface (composition-match strategy on canonical intents).
+- **`mcp/scripts/smoke-register-engine.mjs`** — engine-registry coverage adjusted; still 11/11.
+
+See root [CHANGELOG.md `[Unreleased]`](../../../CHANGELOG.md) for the cross-cutting arc.
+
 ## [0.4.7] - 2026-05-12
 
 ### Changed — smoke + test scripts aligned to post-§72 retrieval surface

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adia-ai/a2ui-mcp",
-  "version": "0.4.7",
+  "version": "0.4.8",
   "description": "AdiaUI A2UI MCP server. Exposes the compose engine over MCP with an engine selector for monolithic + zettel strategies.",
   "type": "module",
   "bin": {

package/scripts/eval-diff.mjs

@@ -71,14 +71,15 @@ if (gateMode === 'combined' && !semanticEnabled) {
   process.exit(2);
 }
 
-if (!['mcp', 'zettel', 'chunk-zettel', 'all'].includes(engine)) {
-  console.error(`[eval-diff] --engine must be one of: mcp | zettel | chunk-zettel | all  (got: ${engine})`);
+if (!['mcp', 'zettel', 'chunk-zettel', 'free-form', 'all'].includes(engine)) {
+  console.error(`[eval-diff] --engine must be one of: mcp | zettel | chunk-zettel | free-form | all  (got: ${engine})`);
   process.exit(2);
 }
 
 const runMcp = engine === 'mcp' || engine === 'all';
 const runZettel = engine === 'zettel' || engine === 'all';
 const runChunkZettel = engine === 'chunk-zettel' || engine === 'all';
+const runFreeForm = engine === 'free-form' || engine === 'all';
 
 // ── MCP adapter: use the top-level patternName exposed by generateInstant ──
 // Shadow-mode capture: when --semantic is set, remember the emitted messages
@@ -119,10 +120,23 @@ async function generateChunkZettelCapture({ intent, mode }) {
   return result;
 }
 
+async function generateFreeFormCapture({ intent }) {
+  // Free-form requires an LLM adapter. generateUI auto-resolves one via
+  // createAdapter — same env-resolution path as monolithic-pro. Without
+  // a key set, the strategy returns `free-form-no-llm` + empty messages
+  // (coverage 0%); that's the honest signal.
+  const result = await generateUI({ intent, engine: 'free-form' });
+  if (semanticEnabled && Array.isArray(result.messages) && result.messages.length > 0) {
+    capturedMessages.set(`free-form:${intent}`, result.messages);
+  }
+  return result;
+}
+
 // ── Run ──
 let mcp = null;
 let zettel = null;
 let chunkZettel = null;
+let freeForm = null;
 
 if (runMcp) {
   console.error(`[eval-diff] running mcp (monolithic) harness…`);
@@ -160,6 +174,18 @@ if (runChunkZettel) {
   console.error(`  coverage=${chunkZettel.coverage}% emitted=${chunkZettel.emitted}/${chunkZettel.total} avgScore=${chunkZettel.avgScoreWhenEmitted}`);
 }
 
+if (runFreeForm) {
+  console.error(`[eval-diff] running free-form (LLM-driven chunk-vocabulary composer) harness…`);
+  freeForm = await runHarnessV2({
+    generate: generateFreeFormCapture,
+    domain,
+    limit,
+    mode: 'instant',
+    label: 'free-form',
+  });
+  console.error(`  coverage=${freeForm.coverage}% emitted=${freeForm.emitted}/${freeForm.total} avgScore=${freeForm.avgScoreWhenEmitted}`);
+}
+
 // ── Shadow-mode semantic validation (Phase 1) ──
 // Opt-in via --semantic. Annotates per-intent rows + aggregates with
 // semanticScore/verdict/combinedScore. DOES NOT affect row.pass, passRate,
@@ -266,6 +292,7 @@ await mkdir(outDir, { recursive: true });
 if (mcp) await writeFile(join(outDir, 'mcp.json'), JSON.stringify(mcp, null, 2));
 if (zettel) await writeFile(join(outDir, 'zettel.json'), JSON.stringify(zettel, null, 2));
 if (chunkZettel) await writeFile(join(outDir, 'chunk-zettel.json'), JSON.stringify(chunkZettel, null, 2));
+if (freeForm) await writeFile(join(outDir, 'free-form.json'), JSON.stringify(freeForm, null, 2));
 
 // ── Build diff.md ──
 function fmt(v) { return v == null ? '—' : String(v); }
@@ -278,10 +305,10 @@ function winner(a, b) {
 }
 
 let md = '';
-md += `# Engine Eval ${[mcp, zettel, chunkZettel].filter(Boolean).length > 1 ? 'Diff' : 'Report'}\n\n`;
+md += `# Engine Eval ${[mcp, zettel, chunkZettel, freeForm].filter(Boolean).length > 1 ? 'Diff' : 'Report'}\n\n`;
 md += `- Run: \`${stamp}\`\n`;
 md += `- Engine(s): ${engine}\n`;
-md += `- Intents: ${(mcp || zettel || chunkZettel).total}${domain ? ` (domain: ${domain})` : ''}${limit ? ` (limit: ${limit})` : ''}\n`;
+md += `- Intents: ${(mcp || zettel || chunkZettel || freeForm).total}${domain ? ` (domain: ${domain})` : ''}${limit ? ` (limit: ${limit})` : ''}\n`;
 md += `- Mode: instant\n`;
 if (semanticEnabled) {
   md += `- Semantic: ${gateMode === 'combined' ? `gating (threshold=${gateThreshold})` : 'shadow'}\n`;
@@ -303,8 +330,8 @@ if (mcp && zettel) {
   }
   md += `| retrieval MRR | ${fmt(mcp.retrievalMRR)} | ${fmt(zettel.retrievalMRR)} |\n\n`;
 } else {
-  const e = mcp || zettel || chunkZettel;
-  const label = mcp ? 'mcp' : zettel ? 'zettel' : 'chunk-zettel';
+  const e = mcp || zettel || chunkZettel || freeForm;
+  const label = mcp ? 'mcp' : zettel ? 'zettel' : chunkZettel ? 'chunk-zettel' : 'free-form';
   md += `| metric | ${label} |\n|---|---:|\n`;
   md += `| coverage % | ${e.coverage} |\n`;
   md += `| emitted | ${e.emitted}/${e.total} |\n`;
@@ -352,8 +379,8 @@ if (mcp && zettel) {
   console.error(`  ties:        ${counts.tie || 0}`);
   console.error(`  both missed: ${counts['both-miss'] || 0}`);
 } else {
-  const e = mcp || zettel || chunkZettel;
-  const label = mcp ? 'mcp' : zettel ? 'zettel' : 'chunk-zettel';
+  const e = mcp || zettel || chunkZettel || freeForm;
+  const label = mcp ? 'mcp' : zettel ? 'zettel' : chunkZettel ? 'chunk-zettel' : 'free-form';
   md += `## Strategy breakdown\n\n`;
   md += `**${label}**: ` + Object.entries(e.strategyBreakdown).map(([k, v]) => `${k}=${v}`).join(', ') + `\n\n`;
   md += `## Per-intent\n\n`;

package/scripts/smoke-engine-registry.mjs

@@ -17,9 +17,11 @@ console.log('[smoke] engines registered:', listEngines().join(', '));
 const monoInstant = pick({ engine: 'monolithic', mode: 'instant' });
 const monoPro = pick({ engine: 'monolithic', mode: 'pro' });
 const zettel = pick({ engine: 'zettel' });
+const freeForm = pick({ engine: 'free-form' });
 console.log('[smoke] pick monolithic/instant:', monoInstant === ENGINES['monolithic-instant'] ? 'ok' : 'FAIL');
 console.log('[smoke] pick monolithic/pro:    ', monoPro === ENGINES['monolithic-pro'] ? 'ok' : 'FAIL');
 console.log('[smoke] pick zettel:            ', zettel === ENGINES.zettel ? 'ok' : 'FAIL');
+console.log('[smoke] pick free-form:         ', freeForm === ENGINES['free-form'] ? 'ok' : 'FAIL');
 console.log('[smoke] pick unknown → fallback:', pick({ engine: 'xxx', mode: 'xxx' }) === ENGINES['monolithic-instant'] ? 'ok' : 'FAIL');
 
 const intent = 'login form with email and password';
@@ -34,10 +36,19 @@ const t2 = Date.now();
 const r2 = await generateUI({ intent, engine: 'zettel' });
 console.log(`[zettel]       ${Date.now() - t2}ms  msgs=${r2.messages?.length}  valid=${r2.validation?.valid}  score=${r2.validation?.score}  strategy=${r2.strategy}  engine=${r2.engine}`);
 
+// Free-form (LLM-driven; runs against the env-resolved adapter via
+// generateUI). With an LLM key → `free-form-composed`; without →
+// `free-form-no-llm`. Smoke verifies the dispatch + shape, not the
+// strategy outcome (which depends on env).
+const t3 = Date.now();
+const r3 = await generateUI({ intent, engine: 'free-form' });
+console.log(`[free-form]    ${Date.now() - t3}ms  msgs=${r3.messages?.length}  strategy=${r3.strategy}  engine=${r3.engine}`);
+
 // Shape invariants
 const ok =
   Array.isArray(r1.messages) && r1.executionId && r1.validation &&
-  Array.isArray(r2.messages) && r2.validation;
+  Array.isArray(r2.messages) && r2.validation &&
+  Array.isArray(r3.messages) && r3.validation && r3.engine === 'free-form';
 console.log(`\n[smoke] shape invariants: ${ok ? 'ok' : 'FAIL'}`);
 
 // Retrieval-quality probe — for each canonical intent, the generated

package/scripts/smoke-register-engine.mjs

@@ -7,8 +7,9 @@ const t = (label, ok, detail = '') => {
   else    { console.log(`  ✗ ${label}  ${detail}`); fail++; }
 };
 
-// Baseline
-t('five built-ins registered', listEngines().length === 5);
+// Baseline — 6 built-ins post-§88 (monolithic-instant, monolithic-pro,
+// monolithic-thinking, zettel, chunk-zettel, free-form).
+t('six built-ins registered', listEngines().length === 6);
 
 // Happy path
 let customCalled = null;