v1/mode: model-aware enrichment downgrade + 3 corpora + variance harness

Pass 5 (5 reps × 4 conditions × 1 file on grok-4.1-fast) showed composing
matrix corpora is anti-additive on strong models — composed lakehouse_arch
+ symbols LOST 5/5 head-to-head vs codereview_isolation (Δ −1.8 grounded
findings, p=0.031). Default flips to isolation; matrix path now auto-
downgrades when the resolved model is strong.

Mode runner:
- matrix_corpus is Vec<String> (string OR array via deserialize_string_or_vec)
- top_k=6 from each corpus, merge by score, take top 8 globally
- chunk tag prefers doc_id over source so reviewer sees [adr:009] vs [lakehouse_arch]
- is_weak_model() gate auto-downgrades codereview_lakehouse → codereview_isolation
  for strong models (default-strong; weak = :free suffix or local last-resort)
- LH_FORCE_FULL_ENRICHMENT=1 bypasses for diagnostic runs
- EnrichmentSources.downgraded_from records when the gate fires

Three corpora indexed via /vectors/index (5849 chunks total):
- lakehouse_arch_v1 — ADRs + phases + PRD + scrum spec (93 docs, 2119 chunks)
- scrum_findings_v1 — past scrum_reviews.jsonl (168 docs, 1260 chunks; EXCLUDED
  from defaults — 24% out-of-bounds line citations from cross-file drift)
- lakehouse_symbols_v1 — regex-extracted pub items + /// docs (656 docs, 2470 chunks)

Experiment infra:
- scripts/build_*_corpus.ts — re-runnable when source content changes
- scripts/mode_pass5_variance_paid.ts — N reps × M conditions on one file
- scripts/mode_pass5_summarize.ts — mean ± σ + head-to-head, parser handles
  numbered + path-with-line + path-with-symbol finding tables
- scripts/mode_compare.ts — groups by mode|corpus when sweeps span corpora
- scripts/mode_experiment.ts — default model bumped to x-ai/grok-4.1-fast,
  --corpus flag for per-call override

Decisions + open follow-ups: docs/MODE_RUNNER_TUNING_PLAN.md

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

scripts/mode_compare.ts

@@ -227,13 +227,37 @@ function main() {
     process.exit(1);
   }
 
-  // Group by file → mode
+  // Group by file → mode (with corpus appended when matrix-bearing modes
+  // were swept across multiple corpora — otherwise lakehouse_arch_v1
+  // would clobber scrum_findings_v1 etc). matrix_corpus is now a Vec
+  // on the wire (multi-corpus support); legacy rows have either a
+  // string or null. Coerce to a stable key.
+  const matrixCorpus = (r: Row): string => {
+    const c = (r.sources as any)?.matrix_corpus;
+    if (!c) return "";
+    if (typeof c === "string") return c;
+    if (Array.isArray(c)) {
+      if (c.length === 0) return "";
+      if (c.length === 1) return c[0];
+      // Stable join: sort then "+"-separate so order doesn't matter.
+      return [...c].sort().join("+");
+    }
+    return "";
+  };
+  const corporaInPlay = new Set(rows.map(matrixCorpus).filter(c => c));
+  const showCorpus = corporaInPlay.size > 1;
+  const keyOf = (r: Row): string => {
+    const c = matrixCorpus(r);
+    return showCorpus && c ? `${r.mode}|${c}` : r.mode;
+  };
+
   const byFile: Record<string, Record<string, Row>> = {};
   const allModes = new Set<string>();
   for (const r of rows) {
     byFile[r.file_path] ??= {};
-    byFile[r.file_path][r.mode] = r; // last-write-wins per mode per file
-    allModes.add(r.mode);
+    const k = keyOf(r);
+    byFile[r.file_path][k] = r; // last-write-wins per (mode,corpus) per file
+    allModes.add(k);
   }
   const modesSorted = [...allModes].sort();
 
@@ -246,13 +270,13 @@ function main() {
     const fileLines = fileContent ? fileContent.split("\n").length : 0;
     console.log(`   (file: ${fileLines} lines${fileContent === null ? ", NOT READABLE — grounding skipped" : ""})`);
     console.log(
-      `  ${pad("mode", 30)} ${pad("resp", 6, true)} ${pad("find", 5, true)} ${pad("ground", 9, true)} ${pad("hallu", 6, true)} ${pad("OOB", 4, true)} ${pad("path", 5, true)} ${pad("ms", 7, true)} ${pad("bug_fp", 6, true)}`
+      `  ${pad("mode", 56)} ${pad("resp", 6, true)} ${pad("find", 5, true)} ${pad("ground", 9, true)} ${pad("hallu", 6, true)} ${pad("OOB", 4, true)} ${pad("path", 5, true)} ${pad("ms", 7, true)} ${pad("bug_fp", 6, true)}`
     );
-    console.log(`  ${"─".repeat(30)} ${"─".repeat(6)} ${"─".repeat(5)} ${"─".repeat(9)} ${"─".repeat(6)} ${"─".repeat(4)} ${"─".repeat(5)} ${"─".repeat(7)} ${"─".repeat(6)}`);
+    console.log(`  ${"─".repeat(56)} ${"─".repeat(6)} ${"─".repeat(5)} ${"─".repeat(9)} ${"─".repeat(6)} ${"─".repeat(4)} ${"─".repeat(5)} ${"─".repeat(7)} ${"─".repeat(6)}`);
     for (const mode of modesSorted) {
       const r = byFile[file][mode];
       if (!r) {
-        console.log(`  ${pad(modeLabel(mode), 30)} ${pad("—", 6, true)}`);
+        console.log(`  ${pad(modeLabel(mode), 56)} ${pad("—", 6, true)}`);
         continue;
       }
       const findings = countFindings(r.response);
@@ -261,7 +285,7 @@ function main() {
       const grounding = checkGrounding(r.response, fileContent);
       const groundedStr = grounding.total === 0 ? "—" : `${grounding.grounded}/${grounding.total}`;
       console.log(
-        `  ${pad(modeLabel(mode), 30)} ${pad(r.response_chars, 6, true)} ${pad(findings, 5, true)} ${pad(groundedStr, 9, true)} ${pad(grounding.hallucinated, 6, true)} ${pad(grounding.out_of_bounds_lines, 4, true)} ${pad(cits, 5, true)} ${pad(r.latency_ms, 7, true)} ${pad(bf, 6, true)}`
+        `  ${pad(modeLabel(mode), 56)} ${pad(r.response_chars, 6, true)} ${pad(findings, 5, true)} ${pad(groundedStr, 9, true)} ${pad(grounding.hallucinated, 6, true)} ${pad(grounding.out_of_bounds_lines, 4, true)} ${pad(cits, 5, true)} ${pad(r.latency_ms, 7, true)} ${pad(bf, 6, true)}`
       );
     }
     console.log("");
@@ -271,11 +295,11 @@ function main() {
   // avg_groundedness is the rate at which findings cite real symbols
   // within file bounds. Modes with low groundedness are confabulating.
   console.log("═══ PER-MODE AGGREGATE ═══\n");
-  console.log(`  ${pad("mode", 30)} ${pad("n", 3, true)} ${pad("avg find", 9, true)} ${pad("avg grnd", 9, true)} ${pad("grnd %", 7, true)} ${pad("avg hallu", 10, true)} ${pad("avg ms", 7, true)}`);
-  console.log(`  ${"─".repeat(30)} ${"─".repeat(3)} ${"─".repeat(9)} ${"─".repeat(9)} ${"─".repeat(7)} ${"─".repeat(10)} ${"─".repeat(7)}`);
+  console.log(`  ${pad("mode", 56)} ${pad("n", 3, true)} ${pad("avg find", 9, true)} ${pad("avg grnd", 9, true)} ${pad("grnd %", 7, true)} ${pad("avg hallu", 10, true)} ${pad("avg ms", 7, true)}`);
+  console.log(`  ${"─".repeat(56)} ${"─".repeat(3)} ${"─".repeat(9)} ${"─".repeat(9)} ${"─".repeat(7)} ${"─".repeat(10)} ${"─".repeat(7)}`);
   const fileCache: Record<string, string | null> = {};
   for (const mode of modesSorted) {
-    const modeRows = rows.filter(r => r.mode === mode);
+    const modeRows = rows.filter(r => keyOf(r) === mode);
     if (modeRows.length === 0) continue;
     const n = modeRows.length;
     let totFind = 0, totGround = 0, totHallu = 0;
@@ -292,7 +316,7 @@ function main() {
     const avgHallu = (totHallu / n).toFixed(1);
     const avgMs = Math.round(modeRows.reduce((s, r) => s + r.latency_ms, 0) / n);
     console.log(
-      `  ${pad(modeLabel(mode), 30)} ${pad(n, 3, true)} ${pad(avgFind, 9, true)} ${pad(avgGround, 9, true)} ${pad(grndPct, 7, true)} ${pad(avgHallu, 10, true)} ${pad(avgMs, 7, true)}`
+      `  ${pad(modeLabel(mode), 56)} ${pad(n, 3, true)} ${pad(avgFind, 9, true)} ${pad(avgGround, 9, true)} ${pad(grndPct, 7, true)} ${pad(avgHallu, 10, true)} ${pad(avgMs, 7, true)}`
     );
   }
 
@@ -301,13 +325,20 @@ function main() {
   // wins). Comparing grounded findings instead corrects for modes
   // that produce convincing-but-fake output.
   console.log("\n═══ MODE vs codereview_lakehouse (grounded findings, per file) ═══\n");
-  console.log(`  ${pad("mode", 30)} ${pad("wins", 5, true)} ${pad("losses", 7, true)} ${pad("ties", 5, true)} ${pad("Δ avg grounded", 16, true)}`);
-  console.log(`  ${"─".repeat(30)} ${"─".repeat(5)} ${"─".repeat(7)} ${"─".repeat(5)} ${"─".repeat(16)}`);
+  console.log(`  ${pad("mode", 56)} ${pad("wins", 5, true)} ${pad("losses", 7, true)} ${pad("ties", 5, true)} ${pad("Δ avg grounded", 16, true)}`);
+  console.log(`  ${"─".repeat(56)} ${"─".repeat(5)} ${"─".repeat(7)} ${"─".repeat(5)} ${"─".repeat(16)}`);
+  // Pick whichever codereview_lakehouse key shows up most often as the
+  // baseline (handles corpus-suffixed keys when showCorpus=true).
+  const baselineKey = modesSorted
+    .filter(k => k.startsWith("codereview_lakehouse"))
+    .sort((a, b) =>
+      Object.values(byFile).filter(f => f[b]).length -
+      Object.values(byFile).filter(f => f[a]).length)[0] ?? "codereview_lakehouse";
   for (const mode of modesSorted) {
-    if (mode === "codereview_lakehouse") continue;
+    if (mode === baselineKey) continue;
     let wins = 0, losses = 0, ties = 0, totalDelta = 0, n = 0;
     for (const file of Object.keys(byFile)) {
-      const baseline = byFile[file]["codereview_lakehouse"];
+      const baseline = byFile[file][baselineKey];
       const challenger = byFile[file][mode];
       if (!baseline || !challenger) continue;
       const fc = fileCache[file] ??= readFileSafe(file);
@@ -322,7 +353,7 @@ function main() {
     if (n === 0) continue;
     const avgDelta = (totalDelta / n).toFixed(1);
     console.log(
-      `  ${pad(modeLabel(mode), 30)} ${pad(wins, 5, true)} ${pad(losses, 7, true)} ${pad(ties, 5, true)} ${pad(avgDelta, 16, true)}`
+      `  ${pad(modeLabel(mode), 56)} ${pad(wins, 5, true)} ${pad(losses, 7, true)} ${pad(ties, 5, true)} ${pad(avgDelta, 16, true)}`
     );
   }
   console.log("\n[compare] done — ⚗ marks lossy/control modes, exclude from recommendations\n");