@mmerterden/multi-agent-pipeline 12.6.0 → 12.7.0

package/pipeline/scripts/smoke-write-state.sh

@@ -93,21 +93,65 @@ else
 fi
 
 # 10. Concurrent writers don't corrupt state
+#
+# Was 5 writers, which lost an update roughly one run in five on a loaded
+# machine and never on an idle one, so it read as a flaky test rather than the
+# data-loss bug it was. 20 writers puts enough writers in the acquire window to
+# make a regression show up on an idle machine too.
 STATE2="$TMPDIR/concurrent.json"
+WRITERS=20
 echo '{"counter":0}' | node "$WRITER" "$STATE2"
-for i in 1 2 3 4 5; do
+for i in $(seq 1 "$WRITERS"); do
   echo "{\"w$i\":$i}" | node "$WRITER" "$STATE2" &
 done
 wait
-# All 5 keys should be present (deep merge)
 MISSING=0
-for i in 1 2 3 4 5; do
+for i in $(seq 1 "$WRITERS"); do
   grep -q "\"w$i\": $i" "$STATE2" || MISSING=$((MISSING+1))
 done
 if [ "$MISSING" -eq 0 ]; then
-  pass "5 concurrent writers all persisted"
+  pass "$WRITERS concurrent writers all persisted"
 else
-  fail "$MISSING concurrent writes lost"
+  fail "$MISSING of $WRITERS concurrent writes lost"
+fi
+
+# 11. A lock with no readable PID must not be treated as stale
+#
+# This is the exact shape of the bug above. The lock used to be created empty
+# and filled on the next line, so a writer that arrived in between read an
+# unparseable PID, called the lock stale, and deleted it while its owner was
+# still working. Both then wrote and one update vanished.
+#
+# An empty lock now has to block until the staleness window elapses, so a
+# writer given a 1s acquire budget must give up with exit 2 rather than help
+# itself to the lock.
+STATE3="$TMPDIR/emptylock.json"
+echo '{"seed":1}' | node "$WRITER" "$STATE3"
+: > "$STATE3.lock"
+LOCK_RC=0
+echo '{"stolen":true}' | WRITE_STATE_LOCK_TIMEOUT_MS=1000 WRITE_STATE_LOCK_STALE_MS=30000 \
+  node "$WRITER" "$STATE3" >/dev/null 2>&1 || LOCK_RC=$?
+if [ "$LOCK_RC" -eq 2 ] && ! grep -q '"stolen"' "$STATE3"; then
+  pass "empty lock is not stolen (exit 2, state untouched)"
+else
+  fail "empty lock was treated as stale (exit $LOCK_RC) - the live-lock race is back"
+fi
+rm -f "$STATE3.lock"
+
+# 12. A lock older than the staleness window is still reclaimed
+#
+# The counterpart to 11: refusing to reclaim an unreadable PID must not mean a
+# genuinely abandoned lock blocks forever.
+STATE4="$TMPDIR/stalelock.json"
+echo '{"seed":1}' | node "$WRITER" "$STATE4"
+: > "$STATE4.lock"
+STALE_RC=0
+echo '{"recovered":true}' | WRITE_STATE_LOCK_TIMEOUT_MS=3000 WRITE_STATE_LOCK_STALE_MS=1 \
+  node "$WRITER" "$STATE4" >/dev/null 2>&1 || STALE_RC=$?
+if [ "$STALE_RC" -eq 0 ] && grep -q '"recovered"' "$STATE4"; then
+  pass "abandoned lock past the staleness window is reclaimed"
+else
+  fail "stale lock not reclaimed (exit $STALE_RC) - writers would wedge"
 fi
 
 echo ""

package/pipeline/scripts/test-gap-rules/android.json

@@ -5,20 +5,20 @@
   "sourceExtensions": [".kt", ".java"],
   "testPathSuffixes": ["Test.kt", "Tests.kt", "Test.java"],
   "testPathContains": ["/src/test/", "/src/androidTest/", "/test/", "/tests/"],
-  "excludePathGlobs": [
-    "build/", ".gradle/", "generated/",
-    "**/BuildConfig.java", "**/R.java"
-  ],
+  "excludePathGlobs": ["build/", ".gradle/", "generated/", "**/BuildConfig.java", "**/R.java"],
   "publicApiPatterns": [
-    {"id": "open_class", "regex": "\\bopen\\s+class\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "open_fun",   "regex": "\\bopen\\s+fun\\s+([a-z_][A-Za-z0-9_]*)"},
-    {"id": "public_fun", "regex": "^\\s*(?:public\\s+)?fun\\s+([a-z_][A-Za-z0-9_]*)"},
-    {"id": "public_class", "regex": "^\\s*(?:public\\s+)?class\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "jvmstatic",  "regex": "@JvmStatic\\b"},
-    {"id": "interface",  "regex": "^\\s*(?:public\\s+)?interface\\s+([A-Z][A-Za-z0-9_]*)"}
+    { "id": "open_class", "regex": "\\bopen\\s+class\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "open_fun", "regex": "\\bopen\\s+fun\\s+([a-z_][A-Za-z0-9_]*)" },
+    { "id": "public_fun", "regex": "^\\s*(?:public\\s+)?fun\\s+([a-z_][A-Za-z0-9_]*)" },
+    { "id": "public_class", "regex": "^\\s*(?:public\\s+)?class\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "jvmstatic", "regex": "@JvmStatic\\b" },
+    { "id": "interface", "regex": "^\\s*(?:public\\s+)?interface\\s+([A-Z][A-Za-z0-9_]*)" }
   ],
   "composablePatterns": [
-    {"id": "composable_fun", "regex": "@Composable\\s+(?:public\\s+|internal\\s+)?fun\\s+([A-Z][A-Za-z0-9_]*)"}
+    {
+      "id": "composable_fun",
+      "regex": "@Composable\\s+(?:public\\s+|internal\\s+)?fun\\s+([A-Z][A-Za-z0-9_]*)"
+    }
   ],
   "expectedTestKindsForComposable": ["compose-ui-test"],
   "hint": "Cover @Composable functions with createComposeRule() + onNodeWithText/Tag/ContentDescription assertions; ViewModels with runTest + Turbine."

package/pipeline/scripts/test-gap-rules/ios.json

@@ -6,22 +6,27 @@
   "testPathSuffixes": ["Tests.swift", "Test.swift"],
   "testPathContains": ["/Tests/", "/UnitTests/", "/SnapshotTests/"],
   "excludePathGlobs": [
-    "Pods/", "Carthage/", ".build/", "DerivedData/",
-    "Generated/", "Sourcery/", "_Generated.swift",
+    "Pods/",
+    "Carthage/",
+    ".build/",
+    "DerivedData/",
+    "Generated/",
+    "Sourcery/",
+    "_Generated.swift",
     "Tests/Fixtures/"
   ],
   "publicApiPatterns": [
-    {"id": "public_func",   "regex": "\\bpublic\\s+func\\s+([A-Za-z_][A-Za-z0-9_]*)"},
-    {"id": "public_class",  "regex": "\\bpublic\\s+(?:final\\s+)?class\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "public_struct", "regex": "\\bpublic\\s+struct\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "public_enum",   "regex": "\\bpublic\\s+enum\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "public_proto",  "regex": "\\bpublic\\s+protocol\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "public_init",   "regex": "\\bpublic\\s+init\\b"},
-    {"id": "objc_export",   "regex": "@objc(?:Members)?\\s+(?:public\\s+)?(?:func|class|var)"}
+    { "id": "public_func", "regex": "\\bpublic\\s+func\\s+([A-Za-z_][A-Za-z0-9_]*)" },
+    { "id": "public_class", "regex": "\\bpublic\\s+(?:final\\s+)?class\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "public_struct", "regex": "\\bpublic\\s+struct\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "public_enum", "regex": "\\bpublic\\s+enum\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "public_proto", "regex": "\\bpublic\\s+protocol\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "public_init", "regex": "\\bpublic\\s+init\\b" },
+    { "id": "objc_export", "regex": "@objc(?:Members)?\\s+(?:public\\s+)?(?:func|class|var)" }
   ],
   "swiftUiViewPatterns": [
-    {"id": "view_struct",     "regex": "struct\\s+([A-Z][A-Za-z0-9_]*)\\s*:\\s*View\\b"},
-    {"id": "config_struct",   "regex": "struct\\s+([A-Z][A-Za-z0-9_]*Configuration)\\b"}
+    { "id": "view_struct", "regex": "struct\\s+([A-Z][A-Za-z0-9_]*)\\s*:\\s*View\\b" },
+    { "id": "config_struct", "regex": "struct\\s+([A-Z][A-Za-z0-9_]*Configuration)\\b" }
   ],
   "expectedTestKindsForView": ["snapshot", "viewinspector", "unit"],
   "snapshotPathHints": ["SnapshotTests", "__Snapshots__"],

package/pipeline/scripts/test-gap-rules/node.json

@@ -3,15 +3,27 @@
   "stack": "node",
   "description": "Node.js / TypeScript test gap rules. Sources in *.ts/*.tsx/*.js/*.mjs; tests in *.test.* / *.spec.* / __tests__/.",
   "sourceExtensions": [".ts", ".tsx", ".js", ".mjs", ".cjs"],
-  "testPathSuffixes": [".test.ts", ".test.tsx", ".test.js", ".test.mjs", ".spec.ts", ".spec.tsx", ".spec.js", ".spec.mjs"],
-  "testPathContains": ["/__tests__/", "/test/", "/tests/"],
-  "excludePathGlobs": [
-    "node_modules/", "dist/", "build/", ".next/", "coverage/",
-    "*.d.ts"
+  "testPathSuffixes": [
+    ".test.ts",
+    ".test.tsx",
+    ".test.js",
+    ".test.mjs",
+    ".spec.ts",
+    ".spec.tsx",
+    ".spec.js",
+    ".spec.mjs"
   ],
+  "testPathContains": ["/__tests__/", "/test/", "/tests/"],
+  "excludePathGlobs": ["node_modules/", "dist/", "build/", ".next/", "coverage/", "*.d.ts"],
   "publicApiPatterns": [
-    {"id": "named_export",   "regex": "^export\\s+(?:async\\s+)?(?:function|class|const|let|interface|type|enum)\\s+([A-Za-z_$][A-Za-z0-9_$]*)"},
-    {"id": "default_export", "regex": "^export\\s+default\\s+(?:async\\s+)?(?:function|class)\\s+([A-Za-z_$][A-Za-z0-9_$]*)"}
+    {
+      "id": "named_export",
+      "regex": "^export\\s+(?:async\\s+)?(?:function|class|const|let|interface|type|enum)\\s+([A-Za-z_$][A-Za-z0-9_$]*)"
+    },
+    {
+      "id": "default_export",
+      "regex": "^export\\s+default\\s+(?:async\\s+)?(?:function|class)\\s+([A-Za-z_$][A-Za-z0-9_$]*)"
+    }
   ],
   "hint": "Cover named exports with vitest/jest test('<name>', ...). React components: React Testing Library (render + screen + userEvent)."
 }

package/pipeline/scripts/test-gap-rules/python.json

@@ -7,13 +7,19 @@
   "testPathPrefixes": ["test_"],
   "testPathContains": ["/tests/", "/test/"],
   "excludePathGlobs": [
-    "__pycache__/", ".venv/", "venv/", ".tox/", "build/", "dist/", "*.egg-info/",
+    "__pycache__/",
+    ".venv/",
+    "venv/",
+    ".tox/",
+    "build/",
+    "dist/",
+    "*.egg-info/",
     "**/migrations/**"
   ],
   "publicApiPatterns": [
-    {"id": "top_def",   "regex": "^def\\s+([a-z_][a-z0-9_]*)"},
-    {"id": "top_class", "regex": "^class\\s+([A-Z][A-Za-z0-9_]*)"},
-    {"id": "async_def", "regex": "^async\\s+def\\s+([a-z_][a-z0-9_]*)"}
+    { "id": "top_def", "regex": "^def\\s+([a-z_][a-z0-9_]*)" },
+    { "id": "top_class", "regex": "^class\\s+([A-Z][A-Za-z0-9_]*)" },
+    { "id": "async_def", "regex": "^async\\s+def\\s+([a-z_][a-z0-9_]*)" }
   ],
   "hint": "Cover with pytest test_<symbol> style. Async with pytest-asyncio. Avoid name collision with private helpers (prefix _)."
 }

package/pipeline/scripts/test-gap-scan.mjs

@@ -90,11 +90,20 @@ function git(args) {
 
 function resolveBase() {
   if (BASE) return BASE;
-  for (const ref of ["origin/main", "main", "origin/master", "master", "develop", "origin/develop"]) {
+  for (const ref of [
+    "origin/main",
+    "main",
+    "origin/master",
+    "master",
+    "develop",
+    "origin/develop",
+  ]) {
     try {
       execFileSync("git", ["-C", ROOT, "rev-parse", "--verify", ref], { stdio: "ignore" });
       return ref;
-    } catch { /* continue */ }
+    } catch {
+      /* continue */
+    }
   }
   throw new Error("no base ref found");
 }
@@ -133,9 +142,15 @@ function isExcluded(path) {
 }
 
 function isTestPath(path) {
-  if (Array.isArray(rules.testPathSuffixes) && rules.testPathSuffixes.some((s) => path.endsWith(s))) return true;
-  if (Array.isArray(rules.testPathPrefixes) && rules.testPathPrefixes.some((p) => basename(path).startsWith(p))) return true;
-  if (Array.isArray(rules.testPathContains) && rules.testPathContains.some((c) => path.includes(c))) return true;
+  if (Array.isArray(rules.testPathSuffixes) && rules.testPathSuffixes.some((s) => path.endsWith(s)))
+    return true;
+  if (
+    Array.isArray(rules.testPathPrefixes) &&
+    rules.testPathPrefixes.some((p) => basename(path).startsWith(p))
+  )
+    return true;
+  if (Array.isArray(rules.testPathContains) && rules.testPathContains.some((c) => path.includes(c)))
+    return true;
   return false;
 }
 
@@ -227,8 +242,11 @@ function scanTreeForTests() {
   if (!SCAN_TREE) return hits;
   const visit = (dir) => {
     let entries;
-    try { entries = readdirSync(dir, { withFileTypes: true }); }
-    catch { return; }
+    try {
+      entries = readdirSync(dir, { withFileTypes: true });
+    } catch {
+      return;
+    }
     for (const ent of entries) {
       const full = join(dir, ent.name);
       const rel = full.replace(ROOT + "/", "");
@@ -238,7 +256,11 @@ function scanTreeForTests() {
         visit(full);
       } else if (ent.isFile() && isTestPath(rel)) {
         let body;
-        try { body = readFileSync(full, "utf8"); } catch { continue; }
+        try {
+          body = readFileSync(full, "utf8");
+        } catch {
+          continue;
+        }
         // crude symbol mention: any Identifier-ish token. Cap body size.
         const slice = body.slice(0, 400_000);
         const tokens = slice.match(/[A-Za-z_][A-Za-z0-9_]{2,}/g) || [];
@@ -258,7 +280,14 @@ function severityFor(symbol, kind) {
   // Public API additions without tests are important; pure boolean exports
   // (e.g. const FLAG = true) tend to be suggestion-tier.
   if (kind === "objc_export" || kind === "interface" || kind === "public_proto") return "important";
-  if (kind.startsWith("public_") || kind === "open_class" || kind === "open_fun" || kind === "named_export" || kind === "default_export") return "suggestion";
+  if (
+    kind.startsWith("public_") ||
+    kind === "open_class" ||
+    kind === "open_fun" ||
+    kind === "named_export" ||
+    kind === "default_export"
+  )
+    return "suggestion";
   if (kind === "view_struct" || kind === "config_struct") return "important";
   if (kind === "composable_fun") return "important";
   if (kind === "top_def" || kind === "top_class" || kind === "async_def") return "suggestion";
@@ -272,7 +301,10 @@ function symbolMentionedInTest(diffText, symbol) {
   let inTestFile = false;
   for (const line of lines) {
     const fm = line.match(/^diff --git a\/(.+?) b\/(.+)$/);
-    if (fm) { inTestFile = isTestPath(fm[2]); continue; }
+    if (fm) {
+      inTestFile = isTestPath(fm[2]);
+      continue;
+    }
     if (!inTestFile) continue;
     if (!line.startsWith("+")) continue;
     if (re.test(line)) return true;
@@ -327,8 +359,8 @@ function main() {
     testsScanned: tests.length,
     gapCount: gaps.length,
     gapBySeverity: {
-      blocking:   gaps.filter((g) => g.severity === "blocking").length,
-      important:  gaps.filter((g) => g.severity === "important").length,
+      blocking: gaps.filter((g) => g.severity === "blocking").length,
+      important: gaps.filter((g) => g.severity === "important").length,
       suggestion: gaps.filter((g) => g.severity === "suggestion").length,
     },
   };

package/pipeline/scripts/test-integrity-gate.mjs

@@ -76,7 +76,11 @@ function main() {
   }
 
   process.stdout.write(
-    JSON.stringify({ schemaVersion: "1.0.0", count: findings.length, findings }, null, PRETTY ? 2 : 0),
+    JSON.stringify(
+      { schemaVersion: "1.0.0", count: findings.length, findings },
+      null,
+      PRETTY ? 2 : 0,
+    ),
   );
 }
 

package/pipeline/scripts/token-budget-report.mjs

@@ -23,10 +23,10 @@ import os from "node:os";
 import path from "node:path";
 
 const args = parseArgs(process.argv.slice(2));
-const METRICS = process.env.METRICS_FILE
-  || path.join(os.homedir(), ".claude/logs/multi-agent/metrics.jsonl");
-const BUDGET = process.env.BUDGET_FILE
-  || path.join(os.homedir(), ".claude/configs/token-budget.json");
+const METRICS =
+  process.env.METRICS_FILE || path.join(os.homedir(), ".claude/logs/multi-agent/metrics.jsonl");
+const BUDGET =
+  process.env.BUDGET_FILE || path.join(os.homedir(), ".claude/configs/token-budget.json");
 const SUMMARY = path.join(os.homedir(), ".claude/logs/multi-agent/metrics-summary.md");
 
 const PERIOD_DAYS = { daily: 1, weekly: 7, monthly: 30 };
@@ -41,17 +41,28 @@ if (!fs.existsSync(METRICS)) {
 
 const lines = fs.readFileSync(METRICS, "utf8").split("\n").filter(Boolean);
 const events = lines
-  .map(l => { try { return JSON.parse(l); } catch { return null; } })
-  .filter(e => e && e.ts)
-  .filter(e => Date.parse(e.ts) >= cutoffMs)
-  .filter(e => !args.task || e.task_id === args.task);
+  .map((l) => {
+    try {
+      return JSON.parse(l);
+    } catch {
+      return null;
+    }
+  })
+  .filter((e) => e && e.ts)
+  .filter((e) => Date.parse(e.ts) >= cutoffMs)
+  .filter((e) => !args.task || e.task_id === args.task);
 
 // Per-task aggregation
 const byTask = new Map();
 for (const e of events) {
-  if (!byTask.has(e.task_id)) byTask.set(e.task_id, {
-    tokens_in: 0, tokens_out: 0, premium_calls: 0, phases: new Set(), models: new Set()
-  });
+  if (!byTask.has(e.task_id))
+    byTask.set(e.task_id, {
+      tokens_in: 0,
+      tokens_out: 0,
+      premium_calls: 0,
+      phases: new Set(),
+      models: new Set(),
+    });
   const t = byTask.get(e.task_id);
   t.phases.add(e.phase);
   const d = e.details || {};
@@ -64,21 +75,26 @@ for (const e of events) {
 // Budget (best-effort)
 let budget = {};
 if (fs.existsSync(BUDGET)) {
-  try { budget = JSON.parse(fs.readFileSync(BUDGET, "utf8")); }
-  catch { /* tolerate malformed */ }
+  try {
+    budget = JSON.parse(fs.readFileSync(BUDGET, "utf8"));
+  } catch {
+    /* tolerate malformed */
+  }
 }
 const limits = {
   perTaskTokens: budget.perTaskTokens || 250000,
   perTaskTokensMultiRepo: budget.perTaskTokensMultiRepo || 500000,
-  perTaskUsd: budget.perTaskUsd || 2.50,
-  perTaskUsdMultiRepo: budget.perTaskUsdMultiRepo || 5.00,
+  perTaskUsd: budget.perTaskUsd || 2.5,
+  perTaskUsdMultiRepo: budget.perTaskUsdMultiRepo || 5.0,
 };
 // Crude USD estimate: opus ~$15/M in, $75/M out; sonnet ~$3/M, $15/M out.
 function estimateUsd(t) {
   const inK = t.tokens_in / 1e6;
   const outK = t.tokens_out / 1e6;
   const opusShare = t.models.has("opus") ? 0.5 : 0; // rough
-  return (inK * (15 * opusShare + 3 * (1 - opusShare))) + (outK * (75 * opusShare + 15 * (1 - opusShare)));
+  return (
+    inK * (15 * opusShare + 3 * (1 - opusShare)) + outK * (75 * opusShare + 15 * (1 - opusShare))
+  );
 }
 
 const summary = {
@@ -100,14 +116,18 @@ for (const [taskId, t] of byTask) {
   const isMulti = t.phases.has("4") && t.premium_calls > 5; // heuristic
   const tokenLimit = isMulti ? limits.perTaskTokensMultiRepo : limits.perTaskTokens;
   const usdLimit = isMulti ? limits.perTaskUsdMultiRepo : limits.perTaskUsd;
-  if (total > tokenLimit) summary.breaches.push({ taskId, kind: "tokens", actual: total, limit: tokenLimit });
-  if (usd > usdLimit)    summary.breaches.push({ taskId, kind: "usd",    actual: usd.toFixed(2), limit: usdLimit });
+  if (total > tokenLimit)
+    summary.breaches.push({ taskId, kind: "tokens", actual: total, limit: tokenLimit });
+  if (usd > usdLimit)
+    summary.breaches.push({ taskId, kind: "usd", actual: usd.toFixed(2), limit: usdLimit });
 }
 
 if (args.json) {
   console.log(JSON.stringify(summary, null, 2));
 } else {
-  console.log(`Token & cost report  -  ${summary.period} (${summary.windowStart} → ${summary.windowEnd})`);
+  console.log(
+    `Token & cost report  -  ${summary.period} (${summary.windowStart} → ${summary.windowEnd})`,
+  );
   console.log(`  tasks:        ${summary.taskCount}`);
   console.log(`  tokens in:    ${summary.totalTokensIn.toLocaleString()}`);
   console.log(`  tokens out:   ${summary.totalTokensOut.toLocaleString()}`);
@@ -123,10 +143,13 @@ if (args.json) {
 }
 
 if (args.append) {
-  const row = `| ${summary.windowEnd.slice(0,10)} | ${summary.period} | ${summary.taskCount} | ${summary.totalTokensIn} | ${summary.totalTokensOut} | $${summary.totalUsd.toFixed(2)} | ${summary.breaches.length} |`;
+  const row = `| ${summary.windowEnd.slice(0, 10)} | ${summary.period} | ${summary.taskCount} | ${summary.totalTokensIn} | ${summary.totalTokensOut} | $${summary.totalUsd.toFixed(2)} | ${summary.breaches.length} |`;
   if (!fs.existsSync(SUMMARY)) {
     fs.mkdirSync(path.dirname(SUMMARY), { recursive: true });
-    fs.writeFileSync(SUMMARY, "# Multi-Agent Pipeline Metrics Summary\n\n| date | period | tasks | tokens_in | tokens_out | est_usd | breaches |\n|---|---|---|---|---|---|---|\n");
+    fs.writeFileSync(
+      SUMMARY,
+      "# Multi-Agent Pipeline Metrics Summary\n\n| date | period | tasks | tokens_in | tokens_out | est_usd | breaches |\n|---|---|---|---|---|---|---|\n",
+    );
   }
   fs.appendFileSync(SUMMARY, row + "\n");
 }