@odavl/guardian 0.2.0 → 1.0.0

package/src/guardian/reality.js

@@ -13,15 +13,34 @@ const { buildAutoAttempts } = require('./auto-attempt-builder');
 const { baselineExists, loadBaseline, saveBaselineAtomic, createBaselineFromSnapshot, compareSnapshots } = require('./baseline-storage');
 const { analyzeMarketImpact, determineExitCodeFromEscalation } = require('./market-criticality');
 const { parsePolicyOption } = require('./preset-loader');
-const { evaluatePolicy } = require('./policy');
+const { evaluatePolicy, loadPolicy } = require('./policy');
 const { aggregateIntelligence } = require('./breakage-intelligence');
 const { writeEnhancedHtml } = require('./enhanced-html-reporter');
-const { printCliSummary } = require('./cli-summary');
+const { analyzePatterns, loadRecentRunsForSite } = require('./pattern-analyzer');
+const crypto = require('crypto');
+const {
+  formatVerdictStatus,
+  formatConfidence,
+  formatVerdictWhy,
+  formatKeyFindings,
+  formatLimits,
+  formatPatternSummary,
+  formatPatternFocus,
+  formatConfidenceDrivers,
+  formatFocusSummary,
+  formatDeltaInsight,
+  // Stage V / Step 5.2: Silence Discipline helpers
+  shouldRenderFocusSummary,
+  shouldRenderDeltaInsight,
+  shouldRenderPatterns,
+  shouldRenderConfidenceDrivers
+} = require('./text-formatters');
 const { sendWebhooks, getWebhookUrl, buildWebhookPayload } = require('./webhook');
 const { findContactOnPage, formatDetectionForReport } = require('./semantic-contact-finder');
 const { formatRunSummary } = require('./run-summary');
 const { isCiMode } = require('./ci-mode');
 const { formatCiSummary, deriveBaselineVerdict } = require('./ci-output');
+const { toCanonicalVerdict, mapExitCodeFromCanonical } = require('./verdicts');
 // Phase 7.1: Performance modes
 const { getTimeoutProfile } = require('./timeout-profiles');
 const { validateAttemptFilter, filterAttempts, filterFlows } = require('./attempts-filter');
@@ -31,12 +50,16 @@ const { executeParallel, validateParallel } = require('./parallel-executor');
 const { BrowserPool } = require('./browser-pool');
 // Phase 7.4: Smart skips
 const { checkPrerequisites } = require('./prerequisite-checker');
-
-function generateRunId(prefix = 'market-run') {
-  const now = new Date();
-  const dateStr = now.toISOString().replace(/[:\-]/g, '').substring(0, 15).replace('T', '-');
-  return `${prefix}-${dateStr}`;
-}
+// Wave 1: Run artifacts naming and metadata
+const { makeRunDirName, makeSiteSlug, writeMetaJson, readMetaJson } = require('./run-artifacts');
+// Wave 2: Latest pointers
+const { updateLatestGlobal, updateLatestBySite } = require('./run-latest');
+// Wave 3: Smart attempt selection
+const { inspectSite, detectProfile } = require('./site-introspection');
+const { filterAttempts: filterAttemptsByRelevance, summarizeIntrospection } = require('./attempt-relevance');
+// Stage II: Golden path
+const { isFirstRun, markFirstRunComplete, applyFirstRunProfile } = require('./first-run-profile');
+const { applyLocalConfig } = require('./config-loader');
 
 function applySafeDefaults(config, warn) {
   const updated = { ...config };
@@ -51,14 +74,30 @@ function applySafeDefaults(config, warn) {
   return updated;
 }
 
+const EXECUTED_OUTCOMES = new Set(['SUCCESS', 'FAILURE', 'FRICTION', 'DISCOVERY_FAILED']);
+const isExecutedAttempt = (attemptResult) => attemptResult && EXECUTED_OUTCOMES.has(attemptResult.outcome);
+const SKIP_CODES = {
+  DISABLED_BY_PRESET: 'DISABLED_BY_PRESET',
+  NOT_APPLICABLE: 'NOT_APPLICABLE',
+  ENGINE_MISSING: 'ENGINE_MISSING',
+  USER_FILTERED: 'USER_FILTERED',
+  PREREQ: 'PREREQUISITE_FAILED'
+};
+
+// Removed compact decision packet writer for Level 1 singleton decision schema
+
 async function executeReality(config) {
   const baseWarn = (...args) => console.warn(...args);
-  const safeConfig = applySafeDefaults(config, baseWarn);
+  const firstRunMode = isFirstRun();
+  // Apply first-run profile if needed (conservative defaults)
+  const profiledConfig = firstRunMode ? applyFirstRunProfile(config) : config;
+  const safeConfig = applySafeDefaults(profiledConfig, baseWarn);
+  const runSignals = [];
 
   const {
     baseUrl,
     attempts = getDefaultAttemptIds(),
-    artifactsDir = './artifacts',
+    artifactsDir = './.odavlguardian',
     headful = false,
     enableTrace = true,
     enableScreenshots = true,
@@ -110,8 +149,11 @@ async function executeReality(config) {
   let filteredAttempts = attempts;
   let filteredFlows = flows;
   if (attemptsFilter && validation && validation.valid && validation.ids.length > 0) {
+    const beforeFilter = Array.isArray(filteredAttempts) ? filteredAttempts.slice() : [];
     filteredAttempts = filterAttempts(attempts, validation.ids);
     filteredFlows = filterFlows(flows, validation.ids);
+    const removed = beforeFilter.filter(id => !filteredAttempts.includes(id));
+    userFilteredAttempts.push(...removed.map(id => ({ attemptId: id, reason: 'Filtered by --attempts' })));
     if (filteredAttempts.length === 0 && filteredFlows.length === 0) {
       console.error('Error: No matching attempts or flows found after filtering');
       console.error(`Hint:  Check your --attempts filter: ${attemptsFilter}`);
@@ -119,6 +161,13 @@ async function executeReality(config) {
     }
   }
 
+  const requestedAttempts = Array.isArray(filteredAttempts) ? filteredAttempts.slice() : [];
+  const disabledByPreset = new Set((config.disabledAttempts || []).map(id => String(id)));
+  const enabledRequestedAttempts = requestedAttempts.filter(id => !disabledByPreset.has(String(id)));
+  const presetDisabledAttempts = requestedAttempts.filter(id => disabledByPreset.has(String(id)));
+  const userFilteredAttempts = [];
+  const missingAttempts = [];
+
   // Phase 7.1: Resolve timeout profile
   const timeoutProfileConfig = getTimeoutProfile(timeoutProfile);
   const resolvedTimeout = timeout || timeoutProfileConfig.default;
@@ -130,11 +179,46 @@ async function executeReality(config) {
     throw new Error(`Invalid URL: ${baseUrl}`);
   }
 
-  const runId = generateRunId();
-  const runDir = path.join(artifactsDir, runId);
+  // Wave 1: Generate human-readable run directory name
+  const startTime = new Date();
+  const siteSlug = makeSiteSlug(baseUrl);
+  // Use 'default' if no policy specified, otherwise extract preset name
+  let policyName = (() => {
+    if (!policy) return 'default';
+    if (typeof policy === 'string') {
+      return policy.startsWith('preset:') ? policy.replace('preset:', '') : policy;
+    }
+    if (typeof policy === 'object' && policy.id) return policy.id;
+    return 'custom';
+  })();
+  // Result will be determined at the end; use placeholder for now
+  let runDirName = makeRunDirName({
+    timestamp: startTime,
+    url: baseUrl,
+    policy: policyName,
+    result: 'PENDING'
+  });
+  let runDir = path.join(artifactsDir, runDirName);
   fs.mkdirSync(runDir, { recursive: true });
+  const runId = runDirName;
   const ciMode = isCiMode();
 
+  // Print positioning message based on policy
+  const isPolicyProtect = policy && ((typeof policy === 'string' && (policy === 'preset:startup' || policy.includes('startup'))) || (typeof policy === 'object' && policy.id === 'startup'));
+  if (!ciMode) {
+    if (isPolicyProtect) {
+      console.log('\nPROTECT MODE: Full market reality test (slower, deeper)');
+    } else {
+      console.log('\nREALITY MODE: Full market reality snapshot');
+    }
+  } else {
+    if (isPolicyProtect) {
+      console.log('PROTECT MODE: Full market reality test');
+    } else {
+      console.log('REALITY MODE: Full market reality snapshot');
+    }
+  }
+
   // Phase 7.1: Print mode info
   if (!ciMode) {
     const modeLines = [];
@@ -152,6 +236,11 @@ async function executeReality(config) {
     console.log(`Base URL: ${baseUrl}`);
     console.log(`Attempts: ${filteredAttempts.join(', ')}`);
     console.log(`Run Dir: ${runDir}`);
+  } else if (firstRunMode) {
+    // Simplified output for first run
+    console.log(`\n🚀 Guardian First Run`);
+    console.log(`━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━`);
+    console.log(`🧪 Scanning: ${baseUrl}`);
   } else {
     console.log(`\n🧪 Market Reality Snapshot v1`);
     console.log(`━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━`);
@@ -168,6 +257,8 @@ async function executeReality(config) {
   let discoveryResult = null;
   let pageLanguage = 'unknown';
   let contactDetectionResult = null;
+  let siteIntrospection = null;
+  let siteProfile = 'unknown';
 
   // Optional: Crawl to discover URLs (lightweight, first N pages)
   if (enableCrawl) {
@@ -177,6 +268,28 @@ async function executeReality(config) {
       await browser.launch(resolvedTimeout);
       await browser.page.goto(baseUrl, { waitUntil: 'domcontentloaded', timeout: resolvedTimeout });
 
+      // Wave 3: Site introspection for smart attempt selection
+      try {
+        console.log(`\n🔬 Inspecting site capabilities...`);
+        siteIntrospection = await inspectSite(browser.page);
+        siteProfile = detectProfile(siteIntrospection);
+        const summary = summarizeIntrospection(siteIntrospection);
+        console.log(`✅ Site profile: ${siteProfile}`);
+        console.log(`   ${summary}`);
+      } catch (introspectionErr) {
+        console.warn(`⚠️  Site introspection failed (non-critical): ${introspectionErr.message}`);
+        // Default to empty introspection if it fails
+        siteIntrospection = {
+          hasLogin: false,
+          hasSignup: false,
+          hasCheckout: false,
+          hasNewsletter: false,
+          hasContactForm: false,
+          hasLanguageSwitch: false
+        };
+        siteProfile = 'unknown';
+      }
+
       // Wave 1.1: Detect page language and contact
       try {
         contactDetectionResult = await findContactOnPage(browser.page, baseUrl);
@@ -194,10 +307,40 @@ async function executeReality(config) {
       await browser.close();
     } catch (crawlErr) {
       console.log(`⚠️  Crawl failed (non-critical): ${crawlErr.message}`);
+      runSignals.push({ id: 'crawl_failed', severity: 'high', type: 'coverage', description: `Crawl failed: ${crawlErr.message}` });
       // Continue anyway - crawl is optional
     }
   }
 
+  // Wave 3: If crawl was disabled but introspection wasn't done, do it now
+  if (!enableCrawl && !siteIntrospection) {
+    console.log(`\n🔬 Inspecting site capabilities...`);
+    const browser = new GuardianBrowser();
+    try {
+      await browser.launch(resolvedTimeout);
+      await browser.page.goto(baseUrl, { waitUntil: 'domcontentloaded', timeout: resolvedTimeout });
+      
+      siteIntrospection = await inspectSite(browser.page);
+      siteProfile = detectProfile(siteIntrospection);
+      const summary = summarizeIntrospection(siteIntrospection);
+      console.log(`✅ Site profile: ${siteProfile}`);
+      console.log(`   ${summary}`);
+      
+      await browser.close();
+    } catch (introspectionErr) {
+      console.warn(`⚠️  Site introspection failed (non-critical): ${introspectionErr.message}`);
+      siteIntrospection = {
+        hasLogin: false,
+        hasSignup: false,
+        hasCheckout: false,
+        hasNewsletter: false,
+        hasContactForm: false,
+        hasLanguageSwitch: false
+      };
+      siteProfile = 'unknown';
+    }
+  }
+
   // Optional: Discovery Engine (Phase 4) — deterministic safe exploration
   if (enableDiscovery) {
     console.log(`\n🔎 Running discovery engine...`);
@@ -217,6 +360,7 @@ async function executeReality(config) {
       await browser.close();
     } catch (discErr) {
       console.log(`⚠️  Discovery failed (non-critical): ${discErr.message}`);
+      runSignals.push({ id: 'discovery_failed', severity: 'high', type: 'discovery', description: `Discovery failed: ${discErr.message}` });
     }
   }
 
@@ -253,7 +397,7 @@ async function executeReality(config) {
   const flowResults = [];
 
   // Determine attempts to run (manual + auto-generated)
-  let attemptsToRun = Array.isArray(attempts) ? attempts.slice() : getDefaultAttemptIds();
+  let attemptsToRun = enabledRequestedAttempts.slice();
   
   // Phase 2: Add auto-generated attempts
   if (enableAutoAttempts && autoAttempts.length > 0) {
@@ -262,11 +406,11 @@ async function executeReality(config) {
     console.log(`➕ Added ${autoAttemptIds.length} auto-generated attempts`);
   }
 
-  if (includeUniversal && !attemptsToRun.includes('universal_reality')) {
+  if (includeUniversal && !attemptsToRun.includes('universal_reality') && !disabledByPreset.has('universal_reality')) {
     attemptsToRun.push('universal_reality');
   }
   // If discovery enabled and site is simple (few interactions), add universal pack
-  if (enableDiscovery && discoveryResult && !attemptsToRun.includes('universal_reality')) {
+  if (enableDiscovery && discoveryResult && !attemptsToRun.includes('universal_reality') && !disabledByPreset.has('universal_reality')) {
     const simpleSite = (discoveryResult.interactionsDiscovered || 0) === 0 || (discoveryResult.pagesVisitedCount || 0) <= 1;
     if (simpleSite) {
       attemptsToRun.push('universal_reality');
@@ -279,6 +423,22 @@ async function executeReality(config) {
     attemptsToRun = filterAttempts(attemptsToRun, validation.ids);
   }
 
+  // Wave 3: Apply smart attempt selection based on introspection
+  let attemptsSkipped = [];
+  if (siteIntrospection) {
+    const attemptObjects = attemptsToRun.map(id => ({ id }));
+    const relevanceResult = filterAttemptsByRelevance(attemptObjects, siteIntrospection);
+    attemptsToRun = relevanceResult.toRun.map(a => a.id);
+    attemptsSkipped = relevanceResult.toSkip;
+    
+    if (attemptsSkipped.length > 0 && !ciMode) {
+      console.log(`\n⊘ Skipping ${attemptsSkipped.length} irrelevant attempt(s):`);
+      for (const skip of attemptsSkipped) {
+        console.log(`    • ${skip.attempt}: ${skip.reason}`);
+      }
+    }
+  }
+
   // Phase 7.2: Print parallel mode if enabled
   if (!ciMode && validatedParallel > 1) {
     console.log(`\n⚡ PARALLEL: ${validatedParallel} concurrent attempts`);
@@ -309,12 +469,27 @@ async function executeReality(config) {
   
   // Phase 7.2: Execute attempts with bounded parallelism
   // Phase 7.3: Pass browser pool to attempts
+  // Phase 7.4: Check applicability before executing
   const attemptResults_parallel = await executeParallel(
     attemptsToRun,
     async (attemptId) => {
       const attemptDef = getAttemptDefinition(attemptId);
       if (!attemptDef) {
-        throw new Error(`Attempt ${attemptId} not found in registry`);
+        missingAttempts.push(attemptId);
+        return {
+          attemptId,
+          attemptName: attemptId,
+          goal: 'Unknown',
+          riskCategory: 'UNKNOWN',
+          source: 'manual',
+          outcome: 'SKIPPED',
+          skipReason: 'Attempt not registered',
+          skipReasonCode: SKIP_CODES.ENGINE_MISSING,
+          exitCode: 0,
+          steps: [],
+          friction: null,
+          error: null
+        };
       }
 
       if (!ciMode) {
@@ -330,8 +505,10 @@ async function executeReality(config) {
 
       let result;
       try {
-        // Phase 7.4: Check prerequisites before executing attempt
+        // Navigate to site to check prerequisites and applicability
         await page.goto(baseUrl, { waitUntil: 'domcontentloaded', timeout: resolvedTimeout });
+        
+        // Phase 7.4: Check prerequisites before executing attempt
         const prereqCheck = await checkPrerequisites(page, attemptId, 2000);
         
         if (!prereqCheck.canProceed) {
@@ -342,27 +519,73 @@ async function executeReality(config) {
           
           result = {
             outcome: 'SKIPPED',
-            skipReason: prereqCheck.reason,
-            exitCode: 0, // SKIPPED does not affect exit code
+              skipReason: prereqCheck.reason,
+              skipReasonCode: SKIP_CODES.PREREQ,
+            exitCode: 0,
             steps: [],
             friction: null,
             error: null
           };
         } else {
-          // Prerequisites met - execute normally
-          result = await executeAttempt({
-            baseUrl,
-            attemptId,
-            artifactsDir: attemptArtifactsDir,
-            headful,
-            enableTrace,
-            enableScreenshots,
-            quiet: ciMode,
-            timeout: resolvedTimeout,
-            // Phase 7.3: Pass context from pool
-            browserContext: context,
-            browserPage: page
-          });
+          // Check if this attempt is applicable to the site
+          const { AttemptEngine } = require('./attempt-engine');
+          const engine = new AttemptEngine({ attemptId, timeout: resolvedTimeout });
+          const applicabilityCheck = await engine.checkAttemptApplicability(page, attemptId);
+          
+          if (!applicabilityCheck.applicable && applicabilityCheck.confidence < 0.3) {
+            // Attempt not applicable - features not present
+            if (!ciMode) {
+              console.log(`    ⊘ Not applicable: ${applicabilityCheck.reason}`);
+            }
+            
+            result = {
+              outcome: 'NOT_APPLICABLE',
+              skipReason: applicabilityCheck.reason,
+              skipReasonCode: SKIP_CODES.NOT_APPLICABLE,
+              discoverySignals: applicabilityCheck.discoverySignals,
+              exitCode: 0,
+              steps: [],
+              friction: null,
+              error: null
+            };
+          } else if (!applicabilityCheck.applicable && applicabilityCheck.confidence >= 0.3 && applicabilityCheck.confidence < 0.6) {
+            // Features possibly present but discovery uncertain - try to execute anyway
+            // Mark as DISCOVERY_FAILED if it fails
+            result = await executeAttempt({
+              baseUrl,
+              attemptId,
+              artifactsDir: attemptArtifactsDir,
+              headful,
+              enableTrace,
+              enableScreenshots,
+              quiet: ciMode,
+              timeout: resolvedTimeout,
+              browserContext: context,
+              browserPage: page
+            });
+            
+            // If execution failed due to element not found, reclassify as DISCOVERY_FAILED
+            if (result.outcome === 'FAILURE' && result.error && /element|selector|locator/i.test(result.error)) {
+              result.outcome = 'DISCOVERY_FAILED';
+              result.skipReason = `Element discovery failed: ${result.error}`;
+              result.discoverySignals = applicabilityCheck.discoverySignals;
+            }
+          } else {
+            // Applicable or high confidence - execute normally
+            result = await executeAttempt({
+              baseUrl,
+              attemptId,
+              artifactsDir: attemptArtifactsDir,
+              headful,
+              enableTrace,
+              enableScreenshots,
+              quiet: ciMode,
+              timeout: resolvedTimeout,
+              // Phase 7.3: Pass context from pool
+              browserContext: context,
+              browserPage: page
+            });
+          }
         }
       } finally {
         // Phase 7.3: Cleanup context after attempt
@@ -378,8 +601,7 @@ async function executeReality(config) {
         ...result
       };
 
-      // Phase 7.1: Fail-fast logic (stop on FAILURE, not FRICTION)
-      // Phase 7.4: SKIPPED does NOT trigger fail-fast
+      // Phase 7.1: Fail-fast logic (stop on FAILURE, not FRICTION or SKIPPED)
       if (failFast && attemptResult.outcome === 'FAILURE') {
         shouldStopScheduling = true;
         if (!ciMode) {
@@ -395,11 +617,79 @@ async function executeReality(config) {
 
   // Collect results in order
   for (const result of attemptResults_parallel) {
-    if (result && !result.skipped) {
+    if (result) {
       attemptResults.push(result);
     }
   }
 
+  // Add explicit SKIPPED entries for attempts filtered out before execution
+  for (const skip of attemptsSkipped) {
+    const def = getAttemptDefinition(skip.attempt) || {};
+    attemptResults.push({
+      attemptId: skip.attempt,
+      attemptName: def.name || skip.attempt,
+      goal: def.goal,
+      riskCategory: def.riskCategory || 'UNKNOWN',
+      source: def.source || 'manual',
+      outcome: 'SKIPPED',
+      skipReason: skip.reason || 'Skipped before execution',
+      skipReasonCode: skip.reasonCode || SKIP_CODES.USER_FILTERED,
+      exitCode: 0,
+      steps: [],
+      friction: null,
+      error: null
+    });
+  }
+
+  // Add explicit SKIPPED entries for attempts disabled by preset (kept for auditability)
+  for (const disabledId of presetDisabledAttempts) {
+    const def = getAttemptDefinition(disabledId) || {};
+    attemptResults.push({
+      attemptId: disabledId,
+      attemptName: def.name || disabledId,
+      goal: def.goal,
+      riskCategory: def.riskCategory || 'UNKNOWN',
+      source: def.source || 'manual',
+      outcome: 'SKIPPED',
+      skipReason: 'Disabled by preset',
+      skipReasonCode: SKIP_CODES.DISABLED_BY_PRESET,
+      exitCode: 0,
+      steps: [],
+      friction: null,
+      error: null,
+      disabledByPreset: true
+    });
+  }
+
+  // Add explicit SKIPPED for user-filtered attempts that were removed before scheduling
+  for (const uf of userFilteredAttempts) {
+    const def = getAttemptDefinition(uf.attemptId) || {};
+    attemptResults.push({
+      attemptId: uf.attemptId,
+      attemptName: def.name || uf.attemptId,
+      goal: def.goal,
+      riskCategory: def.riskCategory || 'UNKNOWN',
+      source: def.source || 'manual',
+      outcome: 'SKIPPED',
+      skipReason: uf.reason,
+      skipReasonCode: SKIP_CODES.USER_FILTERED,
+      exitCode: 0,
+      steps: [],
+      friction: null,
+      error: null,
+      userFiltered: true
+    });
+  }
+
+  // Preserve requested ordering for downstream artifacts
+  const attemptOrder = new Map(requestedAttempts.map((id, idx) => [id, idx]));
+  attemptResults.sort((a, b) => (attemptOrder.get(a.attemptId) ?? 999) - (attemptOrder.get(b.attemptId) ?? 999));
+
+  // Normalize execution metadata
+  for (const result of attemptResults) {
+    result.executed = isExecutedAttempt(result);
+  }
+
   // Phase 3: Execute intent flows (deterministic, curated)
   if (enableFlows) {
     console.log(`\n🎯 Executing intent flows...`);
@@ -526,12 +816,35 @@ async function executeReality(config) {
     });
   }
 
+  // Ensure artifacts exist for skipped attempts so totals remain canonical and inspectable
+  for (const result of attemptResults) {
+    if (!isExecutedAttempt(result)) {
+      const attemptDir = path.join(runDir, result.attemptId);
+      const attemptRunDir = path.join(attemptDir, 'attempt-skipped');
+      fs.mkdirSync(path.join(attemptRunDir, 'attempt-screenshots'), { recursive: true });
+
+      const skipSummary = {
+        attemptId: result.attemptId,
+        outcome: 'SKIPPED',
+        reason: result.skipReason || 'Not executed'
+      };
+
+      const jsonPath = path.join(attemptRunDir, 'attempt-report.json');
+      const htmlPath = path.join(attemptRunDir, 'attempt-report.html');
+      fs.writeFileSync(jsonPath, JSON.stringify(skipSummary, null, 2));
+      fs.writeFileSync(htmlPath, `<html><body><h1>${result.attemptName || result.attemptId}</h1><p>Skipped: ${skipSummary.reason}</p></body></html>`);
+
+      result.reportJsonPath = result.reportJsonPath || jsonPath;
+      result.reportHtmlPath = result.reportHtmlPath || htmlPath;
+    }
+  }
+
   // Generate market report (existing flow)
   const reporter = new MarketReporter();
   const report = reporter.createReport({
     runId,
     baseUrl,
-    attemptsRun: attemptsToRun,
+    attemptsRun: requestedAttempts,
     results: attemptResults.map(r => ({
       attemptId: r.attemptId,
       attemptName: r.attemptName,
@@ -650,22 +963,381 @@ async function executeReality(config) {
     console.warn(`⚠️  Enhanced HTML report failed (non-critical): ${htmlErr.message}`);
   }
 
-  // Phase 5/6: Evaluate policy
+  // Phase 7.3: Cleanup browser pool
+  await browserPool.close();
+
+  // Wave 4: Honest results & near-success
+  // Calculate attempt statistics and coverage
+  const disabledAttemptResults = attemptResults.filter(a => a.disabledByPreset);
+  const eligibleAttempts = attemptResults.filter(a => !a.disabledByPreset);
+  const executedAttempts = eligibleAttempts.filter(isExecutedAttempt);
+  const skippedAttempts = eligibleAttempts.filter(a => !isExecutedAttempt(a));
+  const skippedNotApplicable = eligibleAttempts.filter(a => a.skipReasonCode === SKIP_CODES.NOT_APPLICABLE);
+  const skippedMissing = eligibleAttempts.filter(a => a.skipReasonCode === SKIP_CODES.ENGINE_MISSING);
+  const skippedUserFiltered = eligibleAttempts.filter(a => a.skipReasonCode === SKIP_CODES.USER_FILTERED || a.userFiltered);
+  const skippedDisabledByPreset = disabledAttemptResults;
+  const attemptStats = {
+    total: eligibleAttempts.length,
+    executed: executedAttempts.length,
+    executedCount: executedAttempts.length,
+    enabledPlannedCount: enabledRequestedAttempts.length,
+    disabledPlannedCount: disabledAttemptResults.length,
+    successful: executedAttempts.filter(a => a.outcome === 'SUCCESS').length,
+    failed: executedAttempts.filter(a => a.outcome === 'FAILURE').length,
+    skipped: skippedAttempts.length,
+    skippedDetails: skippedAttempts.map(a => ({ attempt: a.attemptId, reason: a.skipReason || 'Not executed', code: a.skipReasonCode })),
+    disabled: disabledAttemptResults.length,
+    disabledDetails: disabledAttemptResults.map(a => ({ attempt: a.attemptId, reason: a.skipReason || 'Disabled by preset', code: SKIP_CODES.DISABLED_BY_PRESET })),
+    userFiltered: skippedUserFiltered.length,
+    skippedDisabledByPreset: skippedDisabledByPreset.length,
+    skippedNotApplicable: skippedNotApplicable.length,
+    skippedMissing: skippedMissing.length,
+    skippedUserFiltered: skippedUserFiltered.length
+  };
+
+  const nearSuccessDetails = [];
+  for (const a of attemptResults) {
+    if (a.outcome === 'FAILURE') {
+      const errMsg = typeof a.error === 'string' ? a.error : '';
+      const noFailedSteps = Array.isArray(a.steps) ? a.steps.every(s => s.status !== 'failed') : true;
+      if (noFailedSteps && errMsg.includes('Success conditions not met')) {
+        nearSuccessDetails.push({ attempt: a.attemptId, reason: 'Submit succeeded but no confirmation text detected' });
+      }
+    }
+  }
+  attemptStats.nearSuccess = nearSuccessDetails.length;
+  attemptStats.nearSuccessDetails = nearSuccessDetails;
+
+  // Coverage and evidence signals
+  const coverageDenominator = attemptStats.total;
+  const coverageNumerator = attemptStats.executed + skippedNotApplicable.length;
+  const coverageGaps = Math.max(coverageDenominator - coverageNumerator, 0);
+  const coverageSignal = {
+    gaps: coverageGaps,
+    executed: attemptStats.executed,
+    total: coverageDenominator,
+    details: attemptStats.skippedDetails,
+    disabled: attemptStats.disabledDetails,
+    skippedDisabledByPreset: skippedDisabledByPreset.map(a => ({ attempt: a.attemptId, reason: a.skipReason || 'Disabled by preset', code: SKIP_CODES.DISABLED_BY_PRESET })),
+    skippedNotApplicable: skippedNotApplicable.map(a => ({ attempt: a.attemptId, reason: a.skipReason })),
+    skippedMissing: skippedMissing.map(a => ({ attempt: a.attemptId, reason: a.skipReason })),
+    skippedUserFiltered: skippedUserFiltered.map(a => ({ attempt: a.attemptId, reason: a.skipReason })),
+    counts: {
+      executedCount: attemptStats.executed,
+      enabledPlannedCount: attemptStats.enabledPlannedCount,
+      disabledPlannedCount: attemptStats.disabledPlannedCount,
+      skippedDisabledByPreset: skippedDisabledByPreset.length,
+      skippedUserFiltered: skippedUserFiltered.length,
+      skippedNotApplicable: skippedNotApplicable.length,
+      skippedMissing: skippedMissing.length
+    }
+  };
+
+  const evidenceMetrics = {
+    completeness: coverageDenominator > 0 ? coverageNumerator / coverageDenominator : 0,
+    integrity: 0,
+    hashedFiles: 0,
+    totalFiles: 0,
+    screenshotsEnabled: enableScreenshots,
+    tracesEnabled: enableTrace
+  };
+
+  // Build signals used by policy (single source of verdict truth)
+  const policySignals = {
+    coverage: coverageSignal,
+    marketImpact,
+    flows: flowResults,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    attempts: attemptResults,
+    crawlIssues: runSignals.filter(s => s.type === 'coverage'),
+    discoveryIssues: runSignals.filter(s => s.type === 'discovery'),
+    evidence: {
+      metrics: evidenceMetrics,
+      missingScreenshots: !enableScreenshots,
+      missingTraces: !enableTrace
+    },
+    runtimeSignals: runSignals
+  };
+
+  // Resolve policy (strict failure on invalid)
   let policyEval = null;
-  if (policy) {
-    try {
-      const policyObj = parsePolicyOption(policy);
-      if (policyObj) {
-        console.log(`\n🛡️  Evaluating policy...`);
-        policyEval = evaluatePolicy(snapshotBuilder.getSnapshot(), policyObj);
-        console.log(`Policy: ${policyEval.passed ? '✅ PASSED' : '❌ FAILED'}`);
-        if (!policyEval.passed && policyEval.reasons) {
-          policyEval.reasons.slice(0, 3).forEach(r => console.log(`  • ${r}`));
+  let policyObj = null;
+  let presetId = policyName; // Preserve preset ID for naming (e.g., 'startup')
+  let policyHash = null;
+  let policySource = 'default';
+  try {
+    if (policy && typeof policy === 'string' && fs.existsSync(policy)) {
+      policySource = path.resolve(policy);
+    } else if (policy) {
+      policySource = typeof policy === 'string' ? `inline:${policy}` : `preset:${config.preset || policyName}`;
+    } else {
+      policySource = 'default';
+    }
+    policyObj = policy
+      ? (typeof policy === 'object' ? policy : parsePolicyOption(policy))
+      : loadPolicy();
+    policyName = policyObj?.name || policyObj?.id || policyName || (policy ? policy : 'default');
+    policyHash = crypto.createHash('sha256').update(JSON.stringify(policyObj)).digest('hex');
+  } catch (policyLoadErr) {
+    console.error(`Error: ${policyLoadErr.message}`);
+    process.exit(2);
+  }
+
+  // First pass policy evaluation (before manifest integrity)
+  policyEval = evaluatePolicy(snapshotBuilder.getSnapshot(), policyObj, policySignals);
+  console.log(`\n🛡️  Evaluating policy... (${policyName})`);
+  console.log(`Policy evaluation result: exitCode=${policyEval.exitCode}, passed=${policyEval.passed}`);
+  if (policyEval.reasons && policyEval.reasons.length > 0) {
+    policyEval.reasons.slice(0, 3).forEach(r => console.log(`  • ${r}`));
+  }
+
+  const resolvedConfig = {
+    presetId: config.preset || presetId,
+    policySource,
+    policyId: policyObj?.id || policyObj?.name || policyName,
+    policyHash,
+    mediaRequirements: {
+      requireScreenshots: !!(policyObj?.evidence?.requireScreenshots),
+      requireTraces: !!(policyObj?.evidence?.requireTraces),
+      minCompleteness: policyObj?.evidence?.minCompleteness,
+      minIntegrity: policyObj?.evidence?.minIntegrity
+    },
+    attemptPlan: {
+      enabled: enabledRequestedAttempts,
+      disabled: Array.from(disabledByPreset),
+      userFiltered: userFilteredAttempts.map(u => u.attemptId),
+      missing: missingAttempts
+    },
+    coverage: {
+      total: coverageSignal.total,
+      executed: coverageSignal.executed,
+      executedCount: attemptStats.executed,
+      enabledPlannedCount: attemptStats.enabledPlannedCount,
+      disabledPlannedCount: attemptStats.disabledPlannedCount,
+      skippedNotApplicable: coverageSignal.skippedNotApplicable?.length || 0,
+      skippedMissing: coverageSignal.skippedMissing?.length || 0,
+      skippedUserFiltered: coverageSignal.skippedUserFiltered?.length || 0,
+      skippedDisabledByPreset: coverageSignal.skippedDisabledByPreset?.length || 0,
+      disabled: coverageSignal.disabled?.length || 0
+    },
+    evidenceMetrics
+  };
+
+  // Calculate actual duration from start time
+  const endTime = new Date();
+  const actualDurationMs = endTime.getTime() - startTime.getTime();
+
+  // Rename run directory to status placeholder for auditability
+  let exitCode = policyEval.exitCode;
+  const runResultPreManifest = 'PENDING';
+  const priorRunDir = runDir;
+  const finalRunDirName = makeRunDirName({ timestamp: startTime, url: baseUrl, policy: presetId, result: runResultPreManifest });
+  const finalRunDir = path.join(artifactsDir, finalRunDirName);
+  if (finalRunDir !== runDir) {
+    fs.renameSync(runDir, finalRunDir);
+    runDir = finalRunDir;
+    runDirName = finalRunDirName;
+  }
+  // Rebase attempt artifact paths after rename
+  for (const attempt of attemptResults) {
+    if (attempt.attemptJsonPath) {
+      attempt.attemptJsonPath = attempt.attemptJsonPath.replace(priorRunDir, runDir);
+    }
+    if (attempt.stepsLogPath) {
+      attempt.stepsLogPath = attempt.stepsLogPath.replace(priorRunDir, runDir);
+    }
+    if (attempt.reportJsonPath) {
+      attempt.reportJsonPath = attempt.reportJsonPath.replace(priorRunDir, runDir);
+    }
+    if (attempt.reportHtmlPath) {
+      attempt.reportHtmlPath = attempt.reportHtmlPath.replace(priorRunDir, runDir);
+    }
+  }
+  const snapshotPathFinal = path.join(runDir, 'snapshot.json');
+  const marketJsonPathFinal = path.join(runDir, path.basename(jsonPath));
+  const marketHtmlPathFinal = path.join(runDir, path.basename(htmlPath));
+
+  // Build decision artifact and summary (first pass)
+  const initialDecision = computeFinalVerdict({
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    attempts: attemptResults
+  });
+  const initialExplanation = buildRealityExplanation({
+    finalDecision: initialDecision,
+    attemptStats,
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    attempts: attemptResults,
+    coverage: coverageSignal
+  });
+
+  const decisionPath = writeDecisionArtifact({
+    runDir,
+    runId,
+    baseUrl,
+    policyName,
+    preset: config.preset || policyName,
+    finalDecision: initialDecision,
+    attemptStats,
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    resolved: resolvedConfig,
+    attempts: attemptResults,
+    coverage: coverageSignal,
+    explanation: initialExplanation
+  });
+
+  const summaryPath = writeRunSummary(runDir, initialDecision, attemptStats, marketImpact, policyEval, initialExplanation);
+
+  // Build integrity manifest over all artifacts and update evidence metrics
+  try {
+    const manifestInfo = writeIntegrityManifest(runDir);
+    evidenceMetrics.hashedFiles = manifestInfo.hashedFiles;
+    evidenceMetrics.totalFiles = manifestInfo.totalFiles;
+    evidenceMetrics.integrity = manifestInfo.totalFiles > 0 ? manifestInfo.hashedFiles / manifestInfo.totalFiles : 0;
+  } catch (manifestErr) {
+    console.warn(`⚠️  Failed to write integrity manifest: ${manifestErr.message}`);
+    evidenceMetrics.integrity = 0;
+    runSignals.push({ id: 'manifest_failed', severity: 'medium', type: 'evidence', description: `Manifest generation failed: ${manifestErr.message}` });
+  }
+
+  // Re-run policy evaluation with final evidence metrics
+  policyEval = evaluatePolicy(snapshotBuilder.getSnapshot(), policyObj, policySignals);
+  const finalDecision = computeFinalVerdict({
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    attempts: attemptResults
+  });
+  const finalExplanation = buildRealityExplanation({
+    finalDecision,
+    attemptStats,
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    attempts: attemptResults,
+    coverage: coverageSignal
+  });
+  exitCode = finalDecision.exitCode;
+
+  // Rewrite decision + summary with final verdict
+  const decisionPathFinal = writeDecisionArtifact({
+    runDir,
+    runId,
+    baseUrl,
+    policyName,
+    preset: config.preset || policyName,
+    finalDecision,
+    attemptStats,
+    marketImpact,
+    policyEval,
+    baseline: { baselineCreated, baselineSnapshot, diffResult },
+    flows: flowResults,
+    resolved: resolvedConfig,
+    attempts: attemptResults,
+    coverage: coverageSignal,
+    explanation: finalExplanation
+  });
+  const summaryPathFinal = writeRunSummary(runDir, finalDecision, attemptStats, marketImpact, policyEval, finalExplanation);
+
+  const runResult = finalDecision.finalVerdict;
+
+  // Persist policy evaluation and meta into snapshot
+  try {
+    const snap = snapshotBuilder.getSnapshot();
+    snap.policyEvaluation = policyEval;
+    snap.policyName = policyName;
+    snap.meta.policyHash = policyHash;
+    snap.meta.preset = config.preset || presetId;
+    snap.meta.evidenceMetrics = evidenceMetrics;
+    snap.meta.coverage = coverageSignal;
+    snap.meta.resolved = resolvedConfig;
+    snap.resolved = resolvedConfig;
+    snap.meta.result = finalDecision.finalVerdict;
+    snap.meta.attemptsSummary = {
+      executed: attemptStats.executed,
+      successful: attemptStats.successful,
+      failed: attemptStats.failed,
+      skipped: attemptStats.skipped,
+      disabled: attemptStats.disabled,
+      nearSuccess: attemptStats.nearSuccess,
+      nearSuccessDetails: attemptStats.nearSuccessDetails
+    };
+    snap.evidenceMetrics = { ...evidenceMetrics, coverage: coverageSignal };
+    snap.coverage = coverageSignal;
+    await saveSnapshot(snap, snapshotPathFinal);
+    // Minimal attestation: sha256(policyHash + snapshotHash + manifestHash + runId)
+    const snapshotHash = hashFile(snapshotPathFinal);
+    let manifestHash = null;
+    try { manifestHash = hashFile(path.join(runDir, 'manifest.json')); } catch {}
+    const attestationHash = crypto.createHash('sha256').update(`${policyHash}|${snapshotHash}|${manifestHash || 'none'}|${runId}`).digest('hex');
+    snap.meta.attestation = { hash: attestationHash, policyHash, snapshotHash, manifestHash, runId };
+    await saveSnapshot(snap, snapshotPathFinal);
+
+    // Rewrite decision to include attestation and auditor-grade summary
+    writeDecisionArtifact({
+      runDir,
+      runId,
+      baseUrl,
+      policyName,
+      preset: config.preset || policyName,
+      finalDecision,
+      attemptStats,
+      marketImpact,
+      policyEval,
+      baseline: { baselineCreated, baselineSnapshot, diffResult },
+      flows: flowResults,
+      resolved: resolvedConfig,
+      attestation: snap.meta.attestation,
+      audit: {
+        executedAttempts: (snapshotBuilder.getSnapshot()?.attempts || []).filter(a => a.executed).map(a => a.attemptId),
+        notTested: {
+          disabledByPreset: (attemptStats.disabledDetails || []).map(d => d.attempt),
+          userFiltered: (snap.coverage?.skippedUserFiltered || []).map(s => s.attempt),
+          notApplicable: (snap.coverage?.skippedNotApplicable || []).map(s => s.attempt),
+          missing: (snap.coverage?.skippedMissing || []).map(s => s.attempt)
         }
-      }
-    } catch (policyErr) {
-      console.warn(`⚠️  Policy evaluation failed (non-critical): ${policyErr.message}`);
+      },
+      attempts: attemptResults,
+      coverage: coverageSignal,
+      explanation: finalExplanation
+    });
+  } catch (_) {}
+
+  // Persist META.json
+  let metaData;
+  try {
+    metaData = {
+      runDir,
+      url: baseUrl,
+      siteSlug,
+      policy: policyName,
+      policyHash,
+      preset: config.preset || presetId,
+      result: runResult,
+      durationMs: actualDurationMs,
+      profile: siteProfile,
+      attempts: attemptStats,
+      evidence: evidenceMetrics,
+      decisionPath: decisionPathFinal || decisionPath,
+      summaryPath: summaryPathFinal || summaryPath
+    };
+    writeMetaJson(metaData);
+    if (process.env.GUARDIAN_DEBUG) {
+      console.log(`\n💾 META.json written successfully`);
     }
+  } catch (metaErr) {
+    console.warn(`⚠️  Failed to write META.json: ${metaErr.message}`);
+    exitCode = 1;
   }
 
   // Phase 5/6: Send webhook notifications
@@ -688,81 +1360,66 @@ async function executeReality(config) {
     }
   }
 
-  // Determine exit code (including market criticality escalation + policy)
-  let exitCode = 0;
-  const finalSnapshot = snapshotBuilder.getSnapshot();
-
-  if (baselineCreated) {
-    // First run: check market criticality
-    exitCode = 0;
-    if (marketImpact.highestSeverity === 'CRITICAL') {
-      console.log(`🚨 First run with CRITICAL market risks`);
-      exitCode = 1;
-    } else if (marketImpact.highestSeverity === 'WARNING') {
-      console.log(`⚠️  First run with WARNING market risks`);
-      exitCode = 2;
-    }
-    console.log(`✅ Baseline created`);
-  } else if (baselineSnapshot) {
-    // Subsequent runs: check for regressions + severity escalation
-    const baselineSeverity = baselineSnapshot.marketImpactSummary?.highestSeverity || 'INFO';
-    const currentSeverity = marketImpact.highestSeverity;
-    const escalation = determineExitCodeFromEscalation(baselineSeverity, currentSeverity);
-
-    if (escalation.escalated) {
-      // Severity escalation is a FAILURE
-      exitCode = 1;
-      console.log(`🚨 Severity escalated: ${baselineSeverity} → ${currentSeverity}`);
-    } else if (diffResult && Object.keys(diffResult.regressions).length > 0) {
-      exitCode = 1;
-      console.log(`❌ Regressions detected`);
-    } else if (currentSeverity !== 'INFO') {
-      // Still have market risks but didn't escalate
-      exitCode = 2;
-      console.log(`⚠️  ${currentSeverity} market risks present`);
-    } else {
-      exitCode = 0;
-      console.log(`✅ No critical changes`);
+  // Wave 2: Update latest pointers with finalized metadata
+  try {
+    const metaContent = readMetaJson(runDir);
+    updateLatestGlobal(runDir, runDirName, metaContent, artifactsDir);
+    updateLatestBySite(runDir, runDirName, metaContent, artifactsDir);
+    if (process.env.GUARDIAN_DEBUG) {
+      console.log(`✅ Latest pointers updated`);
     }
+  } catch (latestErr) {
+    console.warn(`⚠️  Failed to update latest pointers: ${latestErr.message}`);
   }
 
-  // Override exit code if policy failed
-  if (policyEval && !policyEval.passed) {
-    exitCode = policyEval.exitCode || 1;
-    console.log(`🛡️  Policy override: exit code ${exitCode}`);
-  }
-
-  // Flow-based exit code aggregation (0/1/2)
-  const flowExitCode = computeFlowExitCode(flowResults);
-  exitCode = flowExitCode;
-  if (!ciMode && flowResults.length > 0) {
-    console.log(`Exit code (flows): ${flowExitCode}`);
-  }
-
-  // Phase 7.3: Cleanup browser pool
-  await browserPool.close();
-
   return {
     exitCode,
     report,
     runDir,
-    snapshotPath,
-    marketJsonPath: jsonPath,
-    marketHtmlPath: htmlPath,
+    snapshotPath: snapshotPathFinal,
+    marketJsonPath: marketJsonPathFinal,
+    marketHtmlPath: marketHtmlPathFinal,
     attemptResults,
     flowResults,
     baselineCreated,
     diffResult,
-    snapshot: finalSnapshot,
-    policyEval
+    snapshot: snapshotBuilder.getSnapshot(),
+    policyEval,
+    resolved: resolvedConfig,
+    finalDecision,
+    explanation: finalExplanation,
+    coverage: coverageSignal
   };
 }
 
 async function runRealityCLI(config) {
   try {
-    if (config.watch) {
+    // Stage II: Environment guard (before any other work)
+    const { checkEnvironment, failWithEnvironmentError } = require('./env-guard');
+    const envCheck = checkEnvironment();
+    if (!envCheck.allOk) {
+      failWithEnvironmentError(envCheck.issues);
+      return; // Exit path, but just in case
+    }
+
+    const { config: effectiveConfig, report: configReport } = applyLocalConfig(config);
+
+    console.log('\nConfig:');
+    console.log(`- source: ${configReport.source}`);
+    if (configReport.path) {
+      console.log(`- path: ${configReport.path}`);
+    }
+    console.log('- effective:');
+    console.log(`  - crawl.maxPages: ${configReport.effective.crawl.maxPages}`);
+    console.log(`  - crawl.maxDepth: ${configReport.effective.crawl.maxDepth}`);
+    console.log(`  - timeouts.navigationMs: ${configReport.effective.timeouts.navigationMs}`);
+    console.log(`  - output.dir: ${configReport.effective.output.dir}`);
+
+    const cfg = effectiveConfig;
+
+    if (cfg.watch) {
       const { startWatchMode } = require('./watch-runner');
-      const watchResult = await startWatchMode(config);
+      const watchResult = await startWatchMode(cfg);
       if (watchResult && watchResult.watchStarted === false && typeof watchResult.exitCode === 'number') {
         process.exit(watchResult.exitCode);
       }
@@ -770,10 +1427,30 @@ async function runRealityCLI(config) {
       return;
     }
 
-    const result = await executeReality(config);
+    const result = await executeReality(cfg);
+
+    // Mark first run as complete
+    if (isFirstRun()) {
+      markFirstRunComplete();
+    }
 
     // Phase 6: Print enhanced CLI summary
     const ciMode = isCiMode();
+    if (!ciMode) {
+      const resolved = result.resolved || {};
+      console.log(`\nResolved configuration:`);
+      console.log(`  Preset: ${resolved.presetId || 'unknown'}`);
+      console.log(`  Policy: ${resolved.policyId || 'unknown'} (source: ${resolved.policySource || 'n/a'}, hash: ${resolved.policyHash || 'n/a'})`);
+      if (resolved.mediaRequirements) {
+        console.log(`  Media requirements: screenshots=${resolved.mediaRequirements.requireScreenshots}, traces=${resolved.mediaRequirements.requireTraces}`);
+      }
+      if (resolved.attemptPlan) {
+        console.log(`  Attempt plan: enabled=${(resolved.attemptPlan.enabled || []).length}, disabled=${(resolved.attemptPlan.disabled || []).length}, userFiltered=${(resolved.attemptPlan.userFiltered || []).length}, missing=${(resolved.attemptPlan.missing || []).length}`);
+      }
+      if (resolved.coverage) {
+        console.log(`  Attempt outcomes: executed=${resolved.coverage.executedCount ?? resolved.coverage.executed}, disabled=${resolved.coverage.disabledPlannedCount ?? 0}, skippedDisabledByPreset=${resolved.coverage.skippedDisabledByPreset ?? 0}, skippedUserFiltered=${resolved.coverage.skippedUserFiltered ?? 0}, skippedNotApplicable=${resolved.coverage.skippedNotApplicable ?? 0}, skippedMissing=${resolved.coverage.skippedMissing ?? 0}`);
+      }
+    }
     if (ciMode) {
       const ciSummary = formatCiSummary({
         flowResults: result.flowResults || [],
@@ -784,13 +1461,75 @@ async function runRealityCLI(config) {
       });
       console.log(ciSummary);
     } else {
-      printCliSummary(result.snapshot, result.policyEval);
-      console.log(formatRunSummary({
-        flowResults: result.flowResults || [],
-        diffResult: result.diffResult || null,
-        baselineCreated: result.baselineCreated || false,
-        exitCode: result.exitCode
-      }, { label: 'Summary' }));
+      // Strict CLI summary: traceable, factual only
+      const snap = result.snapshot || {};
+      const meta = snap.meta || {};
+      const coverage = snap.coverage || result.coverage || {};
+      const counts = coverage.counts || {};
+      const evidence = snap.evidenceMetrics || {};
+      const resolved = snap.resolved || {};
+      const finalDecision = result.finalDecision || {};
+      const explanation = result.explanation || buildRealityExplanation({
+        finalDecision,
+        attemptStats: snap.meta?.attemptsSummary || {},
+        marketImpact: snap.marketImpact || {},
+        policyEval: result.policyEval,
+        baseline: { diffResult: result.diffResult },
+        flows: result.flowResults,
+        attempts: result.attemptResults,
+        coverage
+      });
+      const sections = explanation.sections || {};
+      const verdictSection = sections['Final Verdict'] || {};
+      console.log('\n' + '━'.repeat(70));
+      console.log('🛡️  Guardian Reality Summary');
+      console.log('━'.repeat(70) + '\n');
+      console.log(`Target: ${meta.url || 'unknown'}`);
+      console.log(`Run ID: ${meta.runId || 'unknown'}\n`);
+      console.log(`Verdict: ${meta.result || finalDecision.finalVerdict || 'unknown'}`);
+      console.log(`Exit Code: ${result.exitCode}`);
+      if (verdictSection.explanation) console.log(`Reason: ${verdictSection.explanation}`);
+      if (verdictSection.whyNot && verdictSection.whyNot.length > 0) {
+        console.log(`Why not alternatives: ${verdictSection.whyNot.join(' ')}`);
+      }
+      const planned = coverage.total ?? (resolved.coverage?.total);
+      const executed = counts.executedCount ?? (resolved.coverage?.executedCount) ?? coverage.executed;
+      console.log(`Executed / Planned: ${executed} / ${planned}`);
+      const completeness = evidence.completeness ?? resolved.evidenceMetrics?.completeness;
+      const integrity = evidence.integrity ?? resolved.evidenceMetrics?.integrity;
+      console.log(`Coverage Completeness: ${typeof completeness === 'number' ? completeness.toFixed(4) : completeness}`);
+      console.log(`Evidence Integrity: ${typeof integrity === 'number' ? integrity.toFixed(4) : integrity}`);
+      if (meta.attestation?.hash) console.log(`Attestation: ${meta.attestation.hash}`);
+      const executedAttempts = (snap.attempts || []).filter(a => a.executed).map(a => a.attemptId);
+      console.log('\nAudit Summary:');
+      console.log(`  Tested (${executedAttempts.length}): ${executedAttempts.join(', ') || 'none'}`);
+      const skippedDisabled = (coverage.skippedDisabledByPreset || []).map(s => s.attempt);
+      const skippedUserFiltered = (coverage.skippedUserFiltered || []).map(s => s.attempt);
+      const skippedNotApplicable = (coverage.skippedNotApplicable || []).map(s => s.attempt);
+      const skippedMissing = (coverage.skippedMissing || []).map(s => s.attempt);
+      console.log(`  Not Tested — DisabledByPreset (${skippedDisabled.length}): ${skippedDisabled.join(', ') || 'none'}`);
+      console.log(`  Not Tested — UserFiltered (${skippedUserFiltered.length}): ${skippedUserFiltered.join(', ') || 'none'}`);
+      console.log(`  Not Tested — NotApplicable (${skippedNotApplicable.length}): ${skippedNotApplicable.join(', ') || 'none'}`);
+      console.log(`  Not Tested — Missing (${skippedMissing.length}): ${skippedMissing.join(', ') || 'none'}`);
+      if (sections['What Guardian Observed']) {
+        console.log('\nWhat Guardian Observed:');
+        sections['What Guardian Observed'].details.forEach(d => console.log(`  • ${d}`));
+      }
+      if (sections['What Guardian Could Not Confirm']) {
+        console.log('\nWhat Guardian Could Not Confirm:');
+        sections['What Guardian Could Not Confirm'].details.forEach(d => console.log(`  • ${d}`));
+      }
+      if (sections['Evidence Summary']) {
+        console.log('\nEvidence Summary:');
+        sections['Evidence Summary'].details.forEach(d => console.log(`  • ${d}`));
+      }
+      if (sections['Limits of This Run']) {
+        console.log('\nLimits of This Run:');
+        sections['Limits of This Run'].details.forEach(d => console.log(`  • ${d}`));
+      }
+      const reportBase = result.runDir || (configReport?.effective?.output?.dir ? path.join(configReport.effective.output.dir, meta.runId || '') : `artifacts/${meta.runId || ''}`);
+      console.log(`\n📁 Full report: ${reportBase}\n`);
+      console.log('━'.repeat(70));
     }
 
     process.exit(result.exitCode);
@@ -880,4 +1619,426 @@ function computeFlowExitCode(flowResults) {
   return 0;
 }
 
-module.exports = { executeReality, runRealityCLI, computeFlowExitCode, applySafeDefaults };
+function computeFinalVerdict({ marketImpact, policyEval, baseline, flows, attempts }) {
+  const reasons = [];
+
+  const executedAttempts = Array.isArray(attempts)
+    ? attempts.filter(a => a.executed)
+    : [];
+  const successfulAttempts = executedAttempts.filter(a => a.outcome === 'SUCCESS');
+  const failedAttempts = executedAttempts.filter(a => a.outcome === 'FAILURE');
+  const frictionAttempts = executedAttempts.filter(a => a.outcome === 'FRICTION');
+
+  const flowList = Array.isArray(flows) ? flows : [];
+  const failedFlows = flowList.filter(f => f.outcome === 'FAILURE' || f.success === false);
+  const frictionFlows = flowList.filter(f => f.outcome === 'FRICTION');
+  const observedFlows = successfulAttempts.map(a => a.attemptId || a.id || 'unknown');
+
+  // Observation summary always first
+  if (executedAttempts.length > 0) {
+    reasons.push({ code: 'OBSERVED', message: `Observed ${executedAttempts.length} attempted flow(s); successful=${successfulAttempts.length}, failed=${failedAttempts.length}, friction=${frictionAttempts.length}.` });
+  }
+
+  // Baseline regressions
+  const diff = baseline?.diffResult || baseline?.diff;
+  if (diff && diff.regressions && Object.keys(diff.regressions).length > 0) {
+    const regressionAttempts = Object.keys(diff.regressions);
+    reasons.push({ code: 'BASELINE_REGRESSION', message: `Baseline regressions detected for: ${regressionAttempts.join(', ')}.` });
+  }
+
+  // Policy evaluation evidence
+  if (policyEval) {
+    if (!policyEval.passed && policyEval.summary) {
+      reasons.push({ code: 'POLICY', message: policyEval.summary });
+    } else if (!policyEval.passed) {
+      reasons.push({ code: 'POLICY', message: 'Policy conditions not satisfied; evidence insufficient.' });
+    }
+  }
+
+  // Market impact evidence
+  if (marketImpact && marketImpact.highestSeverity) {
+    reasons.push({ code: 'MARKET_IMPACT', message: `Market impact severity observed: ${marketImpact.highestSeverity}.` });
+  }
+
+  // Flow/attempt failures
+  if (failedAttempts.length > 0) {
+    reasons.push({ code: 'CRITICAL_FLOW_FAILURE', message: `Critical flows failed: ${failedAttempts.map(a => a.attemptId || a.id || 'unknown').join(', ')}.` });
+  }
+  if (failedFlows.length > 0) {
+    reasons.push({ code: 'FLOW_FAILURE', message: `Flow executions failed: ${failedFlows.map(f => f.flowId || f.flowName || 'unknown').join(', ')}.` });
+  }
+  if (frictionAttempts.length > 0 || frictionFlows.length > 0) {
+    const frictionIds = [
+      ...frictionAttempts.map(a => a.attemptId || a.id || 'unknown'),
+      ...frictionFlows.map(f => f.flowId || f.flowName || 'unknown')
+    ];
+    reasons.push({ code: 'FLOW_FRICTION', message: `Flows with friction: ${frictionIds.join(', ')}.` });
+  }
+
+  // Determine verdict
+  let internalVerdict;
+  let finalVerdict;
+  let exitCode;
+
+  if (executedAttempts.length === 0 && flowList.length === 0) {
+    internalVerdict = 'INSUFFICIENT_DATA';
+    reasons.unshift({ code: 'NO_OBSERVATIONS', message: 'No meaningful flows executed; only static or configuration checks available.' });
+  } else if (failedAttempts.length === 0 && failedFlows.length === 0 && frictionAttempts.length === 0 && frictionFlows.length === 0 && (!policyEval || policyEval.passed)) {
+    internalVerdict = 'OBSERVED';
+    if (observedFlows.length > 0) {
+      reasons.push({ code: 'OBSERVED_FLOWS', message: `Observed end-to-end flows: ${observedFlows.join(', ')}.` });
+    }
+    reasons.push({ code: 'SCOPE', message: 'Verdict based solely on executed flows; no claim about business readiness beyond observed actions.' });
+  } else {
+    internalVerdict = 'PARTIAL';
+    if (observedFlows.length > 0) {
+      reasons.push({ code: 'PARTIAL_SCOPE', message: `Some flows observed (${observedFlows.join(', ')}), but at least one flow failed or could not be confirmed.` });
+    } else {
+      reasons.push({ code: 'PARTIAL_SCOPE', message: 'Flows were attempted but outcomes include failures or friction; observations are incomplete.' });
+    }
+  }
+
+  // Ensure deterministic ordering: sort by code then message
+  const orderedReasons = reasons
+    .filter(r => r && r.code && r.message)
+    .sort((a, b) => a.code.localeCompare(b.code) || a.message.localeCompare(b.message));
+
+  finalVerdict = toCanonicalVerdict(internalVerdict);
+  exitCode = mapExitCodeFromCanonical(finalVerdict);
+  return { finalVerdict, exitCode, reasons: orderedReasons };
+}
+
+function hashFile(filePath) {
+  const data = fs.readFileSync(filePath);
+  return crypto.createHash('sha256').update(data).digest('hex');
+}
+
+function writeIntegrityManifest(runDir) {
+  const manifest = {
+    generatedAt: new Date().toISOString(),
+    files: []
+  };
+
+  const allFiles = [];
+
+  const walk = (dir) => {
+    const entries = fs.readdirSync(dir, { withFileTypes: true });
+    for (const entry of entries) {
+      const full = path.join(dir, entry.name);
+      if (entry.isDirectory()) {
+        walk(full);
+      } else {
+        allFiles.push(full);
+      }
+    }
+  };
+
+  walk(runDir);
+
+  for (const target of allFiles) {
+    if (fs.existsSync(target)) {
+      manifest.files.push({
+        path: path.relative(runDir, target),
+        sha256: hashFile(target)
+      });
+    }
+  }
+
+  const manifestPath = path.join(runDir, 'manifest.json');
+  fs.writeFileSync(manifestPath, JSON.stringify(manifest, null, 2));
+  return {
+    manifestPath,
+    hashedFiles: manifest.files.length,
+    totalFiles: allFiles.length
+  };
+}
+
+function buildRealityExplanation({ finalDecision = {}, attemptStats = {}, marketImpact = {}, policyEval = {}, baseline = {}, flows = [], attempts = [], coverage = {} }) {
+  const verdict = finalDecision.finalVerdict || 'UNKNOWN';
+  const exitCode = typeof finalDecision.exitCode === 'number' ? finalDecision.exitCode : 1;
+
+  const executedAttempts = (attempts || []).filter(isExecutedAttempt);
+  const successes = executedAttempts.filter(a => a.outcome === 'SUCCESS');
+  const failures = executedAttempts.filter(a => a.outcome === 'FAILURE');
+  const frictions = executedAttempts.filter(a => a.outcome === 'FRICTION');
+
+  const flowList = Array.isArray(flows) ? flows : [];
+  const flowFailures = flowList.filter(f => f.outcome === 'FAILURE' || f.success === false);
+  const flowFrictions = flowList.filter(f => f.outcome === 'FRICTION');
+  const flowSuccesses = flowList.filter(f => f.outcome === 'SUCCESS' || f.success === true);
+
+  const observedDetails = [];
+  if (executedAttempts.length > 0) {
+    const attemptSummary = executedAttempts
+      .map(a => `${a.attemptId || a.id || 'unknown'} (${a.outcome || 'unknown'})`)
+      .sort();
+    observedDetails.push(`Executed ${executedAttempts.length} attempt(s): ${attemptSummary.join(', ')}.`);
+  } else {
+    observedDetails.push('No user journeys executed; evidence limited to crawl/policy signals.');
+  }
+  if (flowSuccesses.length > 0) {
+    const successFlows = flowSuccesses
+      .map(f => f.flowId || f.flowName || 'unknown')
+      .sort();
+    observedDetails.push(`Successful flows: ${successFlows.join(', ')}.`);
+  }
+
+  const couldNotConfirm = [];
+  if (failures.length > 0) {
+    const failureAttempts = failures
+      .map(a => `${a.attemptId || a.id || 'unknown'} (${a.error ? String(a.error).split('\n')[0] : a.outcome})`)
+      .sort();
+    couldNotConfirm.push(`Failures detected in: ${failureAttempts.join(', ')}.`);
+  }
+  if (frictions.length > 0) {
+    const frictionAttempts = frictions.map(a => a.attemptId || a.id || 'unknown').sort();
+    couldNotConfirm.push(`Friction observed in: ${frictionAttempts.join(', ')}.`);
+  }
+  if (flowFailures.length > 0) {
+    const ff = flowFailures.map(f => f.flowId || f.flowName || 'unknown').sort();
+    couldNotConfirm.push(`Flow failures: ${ff.join(', ')}.`);
+  }
+  if (flowFrictions.length > 0) {
+    const ff = flowFrictions.map(f => f.flowId || f.flowName || 'unknown').sort();
+    couldNotConfirm.push(`Flow friction: ${ff.join(', ')}.`);
+  }
+  if (policyEval && policyEval.passed === false) {
+    const reason = policyEval.summary || 'Policy conditions not satisfied.';
+    couldNotConfirm.push(`Policy check failed: ${reason}`);
+  }
+  const diff = baseline?.diffResult || baseline?.diff;
+  if (diff && diff.regressions && Object.keys(diff.regressions).length > 0) {
+    const regressionAttempts = Object.keys(diff.regressions).sort();
+    couldNotConfirm.push(`Baseline regressions: ${regressionAttempts.join(', ')}.`);
+  }
+  if (coverage && typeof coverage.gaps === 'number' && coverage.gaps > 0) {
+    couldNotConfirm.push(`Coverage gaps: ${coverage.gaps} planned attempt(s) not observed.`);
+  }
+  if (couldNotConfirm.length === 0) {
+    couldNotConfirm.push('No outstanding gaps; all planned evidence confirmed for observed scope.');
+  }
+
+  const evidenceSummary = [];
+  const executedCount = attemptStats.executed ?? executedAttempts.length;
+  const totalPlanned = attemptStats.total ?? attemptStats.enabledPlannedCount ?? executedAttempts.length;
+  evidenceSummary.push(`Attempt outcomes: executed=${executedCount}, success=${attemptStats.successful ?? successes.length}, failed=${attemptStats.failed ?? failures.length}, skipped=${attemptStats.skipped ?? 0}.`);
+  evidenceSummary.push(`Flow outcomes: success=${flowSuccesses.length}, failures=${flowFailures.length}, friction=${flowFrictions.length}.`);
+  if (marketImpact && marketImpact.highestSeverity) {
+    evidenceSummary.push(`Market severity observed: ${marketImpact.highestSeverity}.`);
+  } else {
+    evidenceSummary.push('Market severity not reported.');
+  }
+  if (policyEval && typeof policyEval.passed === 'boolean') {
+    evidenceSummary.push(`Policy evaluation: ${policyEval.passed ? 'passed' : 'failed'} (exit ${policyEval.exitCode ?? 'unknown'}).`);
+  } else {
+    evidenceSummary.push('Policy evaluation: not run.');
+  }
+  if (diff && diff.regressions) {
+    const regressions = Object.keys(diff.regressions);
+    evidenceSummary.push(`Baseline regressions: ${regressions.length > 0 ? regressions.join(', ') : 'none'}.`);
+  } else {
+    evidenceSummary.push('Baseline comparison: none or no regressions detected.');
+  }
+
+  const limits = [];
+  const gaps = coverage && typeof coverage.gaps === 'number' ? coverage.gaps : Math.max((totalPlanned || 0) - (executedCount || 0), 0);
+  limits.push(`Coverage: ${executedCount}/${totalPlanned || 0} attempts executed; gaps=${gaps}.`);
+  if (attemptStats.disabled) {
+    limits.push(`Disabled by preset: ${attemptStats.disabled} attempt(s).`);
+  }
+  if (attemptStats.skippedUserFiltered) {
+    limits.push(`User-filtered skips: ${attemptStats.skippedUserFiltered}.`);
+  }
+  if (attemptStats.skippedNotApplicable) {
+    limits.push(`Not-applicable skips: ${attemptStats.skippedNotApplicable}.`);
+  }
+  if (attemptStats.skippedMissing) {
+    limits.push(`Missing engine skips: ${attemptStats.skippedMissing}.`);
+  }
+  if (limits.length === 0) {
+    limits.push('No additional limits detected beyond observed attempts.');
+  }
+
+  let whyThisVerdict;
+  const whyNotList = [];
+  if (verdict === 'OBSERVED') {
+    whyThisVerdict = 'All executed attempts succeeded; no failures, friction, or policy blockers detected.';
+    whyNotList.push('Not PARTIAL because no failures, friction, or policy shortfalls remain.');
+    whyNotList.push(`Not INSUFFICIENT_DATA because ${executedCount} attempt(s) executed with complete evidence.`);
+  } else if (verdict === 'PARTIAL') {
+    const partialDrivers = [];
+    if (failures.length > 0) partialDrivers.push(`${failures.length} failed attempt(s)`);
+    if (frictions.length > 0) partialDrivers.push(`${frictions.length} friction attempt(s)`);
+    if (flowFailures.length > 0) partialDrivers.push(`${flowFailures.length} failed flow(s)`);
+    if (policyEval && policyEval.passed === false) partialDrivers.push('policy not satisfied');
+    whyThisVerdict = partialDrivers.length > 0
+      ? `Evidence is mixed: ${partialDrivers.join('; ')}.`
+      : 'Evidence incomplete or mixed; not all planned signals confirmed.';
+    whyNotList.push('Not OBSERVED because at least one failure, friction, or policy issue remains.');
+    whyNotList.push(`Not INSUFFICIENT_DATA because ${executedCount} attempt(s) executed and produced evidence.`);
+  } else if (verdict === 'INSUFFICIENT_DATA') {
+    whyThisVerdict = 'No meaningful user flows were executed; evidence is insufficient to claim readiness.';
+    whyNotList.push('Not OBSERVED because no successful flows were confirmed.');
+    whyNotList.push('Not PARTIAL because there was no executable evidence to partially support readiness.');
+  } else {
+    whyThisVerdict = 'Verdict unavailable; using conservative interpretation of observed evidence.';
+    whyNotList.push('Alternative verdicts not evaluated due to unknown state.');
+  }
+
+  const finalVerdictSection = {
+    verdict,
+    exitCode,
+    explanation: whyThisVerdict,
+    whyNot: whyNotList,
+    reasons: (finalDecision.reasons || []).map(r => `${r.code}: ${r.message}`)
+  };
+
+  const sections = {
+    'Final Verdict': finalVerdictSection,
+    'What Guardian Observed': {
+      summary: observedDetails[0],
+      details: observedDetails
+    },
+    'What Guardian Could Not Confirm': {
+      summary: couldNotConfirm[0],
+      details: couldNotConfirm
+    },
+    'Evidence Summary': {
+      summary: evidenceSummary[0],
+      details: evidenceSummary
+    },
+    'Limits of This Run': {
+      summary: limits[0],
+      details: limits
+    }
+  };
+
+  return { verdict: finalVerdictSection, sections };
+}
+
+function writeDecisionArtifact({ runDir, runId, baseUrl, policyName, preset, finalDecision, attemptStats, marketImpact, policyEval, baseline, flows, resolved, attestation, audit, attempts = [], coverage = {}, explanation }) {
+  const structuredExplanation = explanation || buildRealityExplanation({ finalDecision, attemptStats, marketImpact, policyEval, baseline, flows, attempts, coverage });
+  const safePolicyEval = policyEval || { passed: true, exitCode: 0, summary: 'Policy evaluation not run.' };
+  const diff = baseline?.diffResult || baseline?.diff || {};
+  const auditSummary = audit ? {
+    tested: Array.isArray(audit.executedAttempts) ? audit.executedAttempts : [],
+    notTested: {
+      disabledByPreset: audit.notTested?.disabledByPreset || [],
+      userFiltered: audit.notTested?.userFiltered || [],
+      notApplicable: audit.notTested?.notApplicable || [],
+      missing: audit.notTested?.missing || []
+    }
+  } : {
+    tested: [],
+    notTested: { disabledByPreset: [], userFiltered: [], notApplicable: [], missing: [] }
+  };
+
+  const decision = {
+    runId,
+    url: baseUrl,
+    timestamp: new Date().toISOString(),
+    preset: preset || 'default',
+    policyName: policyName || 'unknown',
+    finalVerdict: finalDecision.finalVerdict,
+    exitCode: finalDecision.exitCode,
+    reasons: finalDecision.reasons,
+    resolved: resolved || {},
+    attestation: attestation || {},
+    counts: {
+      attemptsExecuted: attemptStats.executed || 0,
+      successful: attemptStats.successful || 0,
+      failed: attemptStats.failed || 0,
+      skipped: attemptStats.skipped || 0,
+      nearSuccess: attemptStats.nearSuccess || 0
+    },
+    inputs: {
+      policy: safePolicyEval,
+      baseline: diff,
+      market: marketImpact || {},
+      flows: {
+        total: Array.isArray(flows) ? flows.length : 0,
+        failures: Array.isArray(flows) ? flows.filter(f => f.outcome === 'FAILURE' || f.success === false).length : 0,
+        frictions: Array.isArray(flows) ? flows.filter(f => f.outcome === 'FRICTION').length : 0
+      }
+    },
+    coverage: {
+      total: coverage.total || attemptStats.enabledPlannedCount || attemptStats.total || 0,
+      executed: coverage.executed || attemptStats.executed || 0,
+      gaps: coverage.gaps ?? Math.max((attemptStats.total || 0) - (attemptStats.executed || 0), 0),
+      skipped: coverage.details || attemptStats.skippedDetails || [],
+      disabled: coverage.disabled || attemptStats.disabledDetails || []
+    },
+    auditSummary,
+    sections: structuredExplanation.sections,
+    explanation: structuredExplanation.verdict
+  };
+
+  const decisionPath = path.join(runDir, 'decision.json');
+  fs.writeFileSync(decisionPath, JSON.stringify(decision, null, 2));
+  return decisionPath;
+}
+
+function writeRunSummary(runDir, finalDecision, attemptStats, marketImpact, policyEval, explanation) {
+  const structuredExplanation = explanation || buildRealityExplanation({ finalDecision, attemptStats, marketImpact, policyEval });
+  const sections = structuredExplanation.sections;
+
+  const lines = [];
+  lines.push(`Final Verdict: ${finalDecision.finalVerdict} (exit ${finalDecision.exitCode})`);
+  lines.push(`Why this verdict: ${sections['Final Verdict'].explanation}`);
+  if (sections['Final Verdict'].whyNot && sections['Final Verdict'].whyNot.length > 0) {
+    lines.push(`Why not alternatives: ${sections['Final Verdict'].whyNot.join(' ')}`);
+  }
+  lines.push('');
+  lines.push('What Guardian Observed:');
+  sections['What Guardian Observed'].details.forEach(d => lines.push(`- ${d}`));
+  lines.push('');
+  lines.push('What Guardian Could Not Confirm:');
+  sections['What Guardian Could Not Confirm'].details.forEach(d => lines.push(`- ${d}`));
+  lines.push('');
+  lines.push('Evidence Summary:');
+  sections['Evidence Summary'].details.forEach(d => lines.push(`- ${d}`));
+  lines.push('');
+  lines.push('Limits of This Run:');
+  sections['Limits of This Run'].details.forEach(d => lines.push(`- ${d}`));
+
+  const summaryPath = path.join(runDir, 'summary.txt');
+  fs.writeFileSync(summaryPath, lines.join('\n'));
+  // Also emit a markdown version for consistency with report discovery
+  try {
+    const mdLines = [];
+    mdLines.push(`# Guardian Reality Summary`);
+    mdLines.push('');
+
+    mdLines.push(`## Final Verdict`);
+    mdLines.push(`- Verdict: ${finalDecision.finalVerdict} (exit ${finalDecision.exitCode})`);
+    mdLines.push(`- Why this verdict: ${sections['Final Verdict'].explanation}`);
+    if (sections['Final Verdict'].whyNot && sections['Final Verdict'].whyNot.length > 0) {
+      sections['Final Verdict'].whyNot.forEach(reason => mdLines.push(`- ${reason}`));
+    }
+    if (sections['Final Verdict'].reasons && sections['Final Verdict'].reasons.length > 0) {
+      mdLines.push(`- Evidence reasons: ${sections['Final Verdict'].reasons.join(' ')}`);
+    }
+
+    mdLines.push('');
+    mdLines.push(`## What Guardian Observed`);
+    sections['What Guardian Observed'].details.forEach(d => mdLines.push(`- ${d}`));
+
+    mdLines.push('');
+    mdLines.push(`## What Guardian Could Not Confirm`);
+    sections['What Guardian Could Not Confirm'].details.forEach(d => mdLines.push(`- ${d}`));
+
+    mdLines.push('');
+    mdLines.push(`## Evidence Summary`);
+    sections['Evidence Summary'].details.forEach(d => mdLines.push(`- ${d}`));
+
+    mdLines.push('');
+    mdLines.push(`## Limits of This Run`);
+    sections['Limits of This Run'].details.forEach(d => mdLines.push(`- ${d}`));
+
+    const summaryMdPath = path.join(runDir, 'summary.md');
+    fs.writeFileSync(summaryMdPath, mdLines.join('\n'));
+  } catch (_) {}
+  return summaryPath;
+}
+
+module.exports = { executeReality, runRealityCLI, computeFlowExitCode, applySafeDefaults, writeDecisionArtifact, computeFinalVerdict };