muaddib-scanner 2.10.82 → 2.10.84

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "muaddib-scanner",
-  "version": "2.10.82",
+  "version": "2.10.84",
   "description": "Supply-chain threat detection & response for npm & PyPI/Python",
   "main": "src/index.js",
   "bin": {

package/src/monitor/ingestion.js

@@ -442,6 +442,10 @@ async function pollNpmChanges(state, scanQueue, stats) {
       // Layer 3: Evaluate if this package should be cached
       const cacheTrigger = evaluateCacheTrigger(name, docMeta, change.doc || null);
 
+      // Layer 2: Extract tarball URL from CouchDB doc (eliminates lazy resolution 404 race)
+      // NOTE: fastTrack flag is computed in resolveTarballAndScan() AFTER metadata
+      // resolution via getNpmLatestTarball(). It cannot be computed here because
+      // post-May 2025, include_docs is deprecated and change.doc is always null.
       scanQueue.push({
         name,
         version: docMeta ? docMeta.version : '',
@@ -643,7 +647,7 @@ async function pollPyPI(state, scanQueue) {
  * @param {Array} scanQueue - Mutable scan queue array
  * @param {Object} stats - Mutable stats object
  */
-const SOFT_BACKPRESSURE_THRESHOLD = 10_000;
+const SOFT_BACKPRESSURE_THRESHOLD = 30_000;
 
 async function poll(state, scanQueue, stats) {
   // Soft backpressure: skip poll when queue is very deep.

package/src/monitor/queue.js

@@ -336,7 +336,7 @@ async function scanPackage(name, version, ecosystem, tarballUrl, registryMeta, s
     let alreadyExtracted = false;
     let extractedDir = null;
 
-    if (unpackedSize > LARGE_PACKAGE_SIZE) {
+    if (unpackedSize > LARGE_PACKAGE_SIZE || meta.fastTrack) {
       // Exception 1: IOC match — always full scan
       let isKnownIOC = false;
       try {
@@ -678,7 +678,10 @@ async function scanPackage(name, version, ecosystem, tarballUrl, registryMeta, s
         stats.suspect++;
 
         // Fire-and-forget tarball archiving — never blocks the pipeline
-        archiveSuspectTarball(name, version, tarballUrl, {
+        // Skip for fast-track packages (large boring enterprise packages — not worth archiving)
+        if (meta.fastTrack) {
+          console.log(`[MONITOR] FAST-TRACK SKIP: ${name}@${version} — skipping archive + LLM (static-only)`);
+        } else archiveSuspectTarball(name, version, tarballUrl, {
           score: riskScore,
           priority: tierLabel,
           rulesTriggered: (result.threats || []).map(t => t.ruleId || t.type).filter(Boolean),
@@ -687,13 +690,35 @@ async function scanPackage(name, version, ecosystem, tarballUrl, registryMeta, s
           console.warn(`[Archive] Failed for ${name}@${version}: ${err.message}`);
         });
 
-        // Sandbox decision based on tier
+        // Sandbox decision based on tier + smart skip for large low-signal packages.
+        // Large packages (>15MB or >80 deps) with only MEDIUM/LOW findings timeout
+        // systematically (90s × 3 = INCONCLUSIVE = 0 detection). Skipping frees slots
+        // for real suspects. Guard-fous: any HIGH/CRITICAL, temporal anomaly, maintainer
+        // change, or dormant spike → sandbox runs regardless of size.
+        const SANDBOX_SIZE_SKIP_BYTES = 15 * 1024 * 1024; // 15MB
+        const SANDBOX_DEPS_SKIP = 80;
+        const isLargePackage = (meta.unpackedSize || 0) > SANDBOX_SIZE_SKIP_BYTES ||
+          (meta.dependencyCount || 0) > SANDBOX_DEPS_SKIP;
+        const hasHighOrCriticalFinding = (result.summary.critical || 0) > 0 || (result.summary.high || 0) > 0;
+        const hasTemporalSignal = (result.threats || []).some(t =>
+          t.type === 'postinstall_added' || t.type === 'preinstall_added' ||
+          t.type === 'install_added' || t.type === 'maintainer_change' ||
+          t.type === 'dormant_spike' || t.type === 'publish_anomaly'
+        );
+        const skipSandboxLargePackage = (isLargePackage || meta.fastTrack) && !hasHighOrCriticalFinding && !hasTemporalSignal;
+
+        if (skipSandboxLargePackage && meta.fastTrack) {
+          console.log(`[MONITOR] FAST-TRACK: ${name}@${version} — large package static-only (${((meta.unpackedSize || 0) / 1024 / 1024).toFixed(1)}MB, no lifecycle scripts)`);
+        } else if (skipSandboxLargePackage) {
+          console.log(`[MONITOR] SANDBOX SKIP (large low-signal): ${name}@${version} (${((meta.unpackedSize || 0) / 1024 / 1024).toFixed(1)}MB, deps=${meta.dependencyCount || '?'}, no HIGH/CRIT, no temporal)`);
+        }
+
         // T1a: mandatory sandbox (HC malice types, TIER1_TYPES non-LOW, lifecycle + intent compound)
         // T1b: conditional sandbox (HIGH/CRITICAL without HC type — bundler FP zone)
         //       → sandbox only if score >= 25 (significant risk) or queue pressure is low
         // T2: sandbox if queue < 50 (as before)
         let sandboxResult = null;
-        const shouldSandbox = isSandboxEnabled() && sandboxAvailable && (
+        const shouldSandbox = !skipSandboxLargePackage && isSandboxEnabled() && sandboxAvailable && (
           tier === '1a' ||
           (tier === '1b' && (riskScore >= 25 || scanQueue.length < 20)) ||
           (tier === 2 && scanQueue.length < 50)
@@ -845,8 +870,9 @@ async function scanPackage(name, version, ecosystem, tarballUrl, registryMeta, s
         // Record daily alert with post-reputation score for top suspects ranking
         dailyAlerts.push({ name, version, ecosystem, findingsCount: result.summary.total, score: adjustedResult.summary.riskScore || 0, tier });
         // LLM Detective: AI-powered analysis for T1a/T1b suspects
+        // Skip for fast-track (large boring packages — LLM analysis adds 10-30s for no value)
         let llmResult = null;
-        if ((tier === '1a' || tier === '1b') && (adjustedResult.summary.riskScore || 0) >= 25) {
+        if (!meta.fastTrack && (tier === '1a' || tier === '1b') && (adjustedResult.summary.riskScore || 0) >= 25) {
           try {
             const { investigatePackage, isLlmEnabled, getLlmMode } = require('../ml/llm-detective.js');
             if (isLlmEnabled()) {
@@ -1057,6 +1083,19 @@ async function resolveTarballAndScan(item, stats, dailyAlerts, recentlyScanned,
       if (npmInfo.version) item.version = npmInfo.version;
       if (npmInfo.unpackedSize) item.unpackedSize = npmInfo.unpackedSize;
       if (npmInfo.scripts) item.registryScripts = npmInfo.scripts;
+
+      // Fast-track decision: large packages (>15MB) with no lifecycle scripts and no IOC match.
+      // Computed HERE (after metadata resolution), not at ingestion time — post-May 2025
+      // CouchDB changes feed has no docs, so metadata is only available after lazy fetch.
+      // Fast-track packages get: quick static scan (package.json + shell only), no AST,
+      // no sandbox, no LLM, no archiving. Exits in ~2-3s instead of 30-300s.
+      const FAST_TRACK_SIZE_BYTES = 15 * 1024 * 1024;
+      if (!item.isIOCMatch && (item.unpackedSize || 0) > FAST_TRACK_SIZE_BYTES) {
+        const scripts = item.registryScripts || {};
+        if (!scripts.preinstall && !scripts.postinstall && !scripts.install) {
+          item.fastTrack = true;
+        }
+      }
     } catch (err) {
       console.error(`[MONITOR] ERROR resolving npm tarball for ${item.name}: ${err.message}`);
       recordError(err, stats);
@@ -1114,9 +1153,11 @@ async function resolveTarballAndScan(item, stats, dailyAlerts, recentlyScanned,
   let publishResult = null;
   let maintainerResult = null;
 
-  if (item.ecosystem === 'npm') {
+  if (item.ecosystem === 'npm' && !item.fastTrack) {
     // Run all 4 temporal checks in parallel — each is independent.
     // With metadata cache (temporal-analysis.js), the 4 modules share 1 HTTP request.
+    // Skipped for fast-track packages (large boring packages — temporal checks make
+    // 4 HTTP requests to npm registry per package, pointless for 50MB enterprise packages).
     const [tempRes, astRes, pubRes, maintRes] = await Promise.allSettled([
       runTemporalCheck(item.name, dailyAlerts),
       runTemporalAstCheck(item.name, dailyAlerts),
@@ -1135,7 +1176,8 @@ async function resolveTarballAndScan(item, stats, dailyAlerts, recentlyScanned,
   const scanResult = await scanPackage(item.name, item.version, item.ecosystem, item.tarballUrl, {
     unpackedSize: item.unpackedSize || 0,
     registryScripts: item.registryScripts || null,
-    _cacheTrigger: item._cacheTrigger || null
+    _cacheTrigger: item._cacheTrigger || null,
+    fastTrack: item.fastTrack || false
   }, stats, dailyAlerts, recentlyScanned, downloadsCache, scanQueue, sandboxAvailable);
   const sandboxResult = scanResult && scanResult.sandboxResult;
   const staticClean = scanResult && scanResult.staticClean;

package/src/sandbox/index.js

@@ -224,7 +224,12 @@ async function runSingleSandbox(packageName, options = {}) {
       '--cap-drop=ALL'
     ];
 
-    // gVisor runtime: use runsc instead of default runc
+    // gVisor runtime: use runsc instead of default runc.
+    // Performance: configure --directfs and --overlay2=all:memory in daemon.json:
+    //   "runsc": { "path": "/usr/bin/runsc", "runtimeArgs": ["--directfs", "--overlay2=all:memory"] }
+    // --directfs: bypass gofer process for direct filesystem access (fewer RPCs, faster I/O)
+    // --overlay2=all:memory: sandbox writes go to tmpfs instead of host (faster, isolated)
+    // These flags require gVisor >= 2023-06-01.
     if (gvisorMode) {
       dockerArgs.push('--runtime=runsc');
       dockerArgs.push('-e', 'MUADDIB_GVISOR=1');