agentic-qe 3.15.9 → 3.15.10

package/dist/cli/chunks/{workflow-QMZPAWQW.js → workflow-EOVVCN2E.js}

@@ -1,5 +1,5 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.15.9");process.exit(0)}
-import{b as $}from"./chunk-6EPHNJTL.js";import"./chunk-KIKWJM6J.js";import{a as k,b as W,c as b,g as C}from"./chunk-WGDR2KIR.js";import"./chunk-7QGYJ5OY.js";import{c as v}from"./chunk-PVIA2FPD.js";import"./chunk-JNNZK7KH.js";import"./chunk-N3KEY4XJ.js";import"./chunk-HZ5NC6U5.js";import{Command as D}from"commander";import o from"chalk";function O(f,i){let h=f.indexOf("=");return h>0&&(i[f.substring(0,h)]=f.substring(h+1)),i}function V(f,i,h){let m=new D("workflow").description("Manage QE workflows and pipelines (ADR-041)");return m.command("run <file>").description("Execute a QE pipeline from YAML file").option("-w, --watch","Watch execution progress").option("-v, --verbose","Show detailed output").option("--params <json>","Additional parameters as JSON","{}").action(async(n,t)=>{if(!await h())return;let c=(await import("path")).resolve(n);try{console.log(o.blue(`
+import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.15.10");process.exit(0)}
+import{b as $}from"./chunk-ZXPY654C.js";import"./chunk-M2QV7PT7.js";import{a as k,b as W,c as b,g as C}from"./chunk-4G63WT4F.js";import"./chunk-FZOWULMH.js";import{c as v}from"./chunk-ACWFKES2.js";import"./chunk-NSHU347M.js";import"./chunk-GHBNRMSJ.js";import"./chunk-OY2AHLSF.js";import{Command as D}from"commander";import o from"chalk";function O(f,i){let h=f.indexOf("=");return h>0&&(i[f.substring(0,h)]=f.substring(h+1)),i}function V(f,i,h){let m=new D("workflow").description("Manage QE workflows and pipelines (ADR-041)");return m.command("run <file>").description("Execute a QE pipeline from YAML file").option("-w, --watch","Watch execution progress").option("-v, --verbose","Show detailed output").option("--params <json>","Additional parameters as JSON","{}").action(async(n,t)=>{if(!await h())return;let c=(await import("path")).resolve(n);try{console.log(o.blue(`
  Running workflow from: ${n}
 `));let l=k(c);if(!l.success||!l.workflow){console.log(o.red("Failed to parse pipeline:"));for(let d of l.errors)console.log(o.red(`   ${d}`));await i(1)}let u={...v(t.params,"params")};if(l.pipeline){for(let d of l.pipeline.stages)if(d.params)for(let[y,p]of Object.entries(d.params))u[y]=p}if(!f.workflowOrchestrator.getWorkflow(l.workflow.id)){let d=f.workflowOrchestrator.registerWorkflow(l.workflow);d.success||(console.log(o.red(`Failed to register workflow: ${d.error.message}`)),await i(1))}let g=await f.workflowOrchestrator.executeWorkflow(l.workflow.id,u);if(!g.success){console.log(o.red(`Failed to start workflow: ${g.error.message}`)),await i(1);return}let a=g.value;if(console.log(o.cyan(`  Execution ID: ${a}`)),console.log(o.gray(`  Workflow: ${l.workflow.name}`)),console.log(o.gray(`  Stages: ${l.workflow.steps.length}`)),console.log(""),t.watch){console.log(o.blue(`Workflow Progress:
 `));let d,y=Date.now();for(;;){let w=f.workflowOrchestrator.getWorkflowStatus(a);if(!w)break;if(!d||d.progress!==w.progress||d.status!==w.status||JSON.stringify(d.currentSteps)!==JSON.stringify(w.currentSteps)){process.stdout.write("\r\x1B[K");let S="\u2588".repeat(Math.floor(w.progress/5))+"\u2591".repeat(20-Math.floor(w.progress/5)),R=w.status==="completed"?o.green:w.status==="failed"?o.red:w.status==="running"?o.yellow:o.gray;console.log(`  [${S}] ${w.progress}% - ${R(w.status)}`),w.currentSteps.length>0&&t.verbose&&console.log(o.gray(`    Running: ${w.currentSteps.join(", ")}`)),d=w}if(w.status==="completed"||w.status==="failed"||w.status==="cancelled")break;await new Promise(S=>setTimeout(S,500))}let p=f.workflowOrchestrator.getWorkflowStatus(a);if(p){console.log("");let w=p.duration||Date.now()-y;p.status==="completed"?(console.log(o.green("Workflow completed successfully")),console.log(o.gray(`   Duration: ${$(w)}`)),console.log(o.gray(`   Completed: ${p.completedSteps.length} stages`)),p.skippedSteps.length>0&&console.log(o.yellow(`   Skipped: ${p.skippedSteps.length} stages`))):p.status==="failed"?(console.log(o.red("Workflow failed")),console.log(o.red(`   Error: ${p.error}`)),console.log(o.gray(`   Failed stages: ${p.failedSteps.join(", ")}`))):console.log(o.yellow(`Workflow ${p.status}`))}console.log(""),await i(p?.status==="completed"?0:1);return}else console.log(o.green("Workflow execution started")),console.log(o.gray(`   Use 'aqe workflow status ${a}' to check progress`));console.log(""),await i(0)}catch(l){console.error(o.red(`
@@ -35,10 +35,10 @@ Workflow cancelled: ${n}
 `)):console.log(o.red(`
 Failed to cancel workflow: ${t.error.message}
 `)),await i(t.success?0:1)}catch(t){console.error(o.red(`
-Failed to cancel workflow:`),t),await i(1)}}),m.command("browser-list").description("List available browser workflow templates").action(async()=>{try{let{BrowserWorkflowTool:n}=await import("./browser-workflow-NDXRROWZ.js"),e=await new n().invoke({});if(e.success&&e.data){console.log(o.blue(`
+Failed to cancel workflow:`),t),await i(1)}}),m.command("browser-list").description("List available browser workflow templates").action(async()=>{try{let{BrowserWorkflowTool:n}=await import("./browser-workflow-MSBOLYUY.js"),e=await new n().invoke({});if(e.success&&e.data){console.log(o.blue(`
  Browser Workflow Templates:
 `));for(let c of e.data.availableTemplates)console.log(`  ${o.cyan(c)}`);console.log("")}else console.log(o.red(`Failed: ${e.error||"Unknown error"}`));await i(0)}catch(n){console.error(o.red(`
-Failed:`),n),await i(1)}}),m.command("browser-load [template]").description("Load and validate a browser workflow template or inline YAML").option("--yaml <yaml>","Inline YAML workflow definition").option("-v, --var <key=value>","Variable override (repeatable)",O,{}).option("-F, --format <format>","Output format (text|json)","text").option("-o, --output <path>","Write output to file").action(async(n,t)=>{try{let{BrowserWorkflowTool:e}=await import("./browser-workflow-NDXRROWZ.js"),{writeOutput:c,toJSON:l}=await import("./ci-output-M47NJY6D.js"),s=new e,u={variables:t.var||{}};if(t.yaml)u.workflowYaml=t.yaml;else if(n)if(n.endsWith(".yaml")||n.endsWith(".yml")){let g=await import("fs"),d=(await import("path")).resolve(n);g.existsSync(d)||(console.log(o.red(`
+Failed:`),n),await i(1)}}),m.command("browser-load [template]").description("Load and validate a browser workflow template or inline YAML").option("--yaml <yaml>","Inline YAML workflow definition").option("-v, --var <key=value>","Variable override (repeatable)",O,{}).option("-F, --format <format>","Output format (text|json)","text").option("-o, --output <path>","Write output to file").action(async(n,t)=>{try{let{BrowserWorkflowTool:e}=await import("./browser-workflow-MSBOLYUY.js"),{writeOutput:c,toJSON:l}=await import("./ci-output-FXZ7ZZW3.js"),s=new e,u={variables:t.var||{}};if(t.yaml)u.workflowYaml=t.yaml;else if(n)if(n.endsWith(".yaml")||n.endsWith(".yml")){let g=await import("fs"),d=(await import("path")).resolve(n);g.existsSync(d)||(console.log(o.red(`
 File not found: ${d}
 `)),await i(1)),u.workflowYaml=g.readFileSync(d,"utf-8")}else u.templateName=n;else console.log(o.red(`
 Provide a template name or --yaml. Use "workflow browser-list" to see templates.

package/dist/cli/chunks/workflow-orchestrator-UKHEJVSK.js

@@ -0,0 +1,2 @@
+import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.15.10");process.exit(0)}
+import{a,b,d as c,e as d}from"./chunk-MA7JIQ6U.js";import"./chunk-LRZY4KS2.js";import"./chunk-C522H7XR.js";import"./chunk-3Y7SPD72.js";import"./chunk-ACWFKES2.js";import"./chunk-AI4VYSFS.js";import"./chunk-VJQYDYAJ.js";import"./chunk-OBAREPDC.js";import"./chunk-6GP6W46M.js";import"./chunk-PGYMKGOE.js";import"./chunk-GZLZCZ42.js";import"./chunk-GJMLK5UT.js";import"./chunk-6PX4OXQO.js";import"./chunk-WQ4PYWSW.js";import"./chunk-NSHU347M.js";import"./chunk-AH3ZD3NJ.js";import"./chunk-HOZUDE4C.js";import"./chunk-ATHWDAXB.js";import"./chunk-N6PIIK6G.js";import"./chunk-WSF24I3D.js";import"./chunk-GHBNRMSJ.js";import"./chunk-XLZYRTNO.js";import"./chunk-EWFECZ2U.js";import"./chunk-VWNFAW2M.js";import"./chunk-XBVUVFJK.js";import"./chunk-EN2QSKFF.js";import"./chunk-OY2AHLSF.js";export{b as DEFAULT_WORKFLOW_CONFIG,a as WorkflowEvents,c as WorkflowOrchestrator,d as createWorkflowOrchestrator};

package/dist/cli/chunks/wrappers-AGODHBOO.js

@@ -0,0 +1,2 @@
+import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.15.10");process.exit(0)}
+import{a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v as L,w as M,x as N}from"./chunk-ODJNUMV3.js";import{a as v,b as w,c as x,d as y,e as z,f as A,g as B,h as C,i as D,j as E,k as F,l as G,m as H,n as I,o as J,p as K}from"./chunk-ESUO3GTJ.js";import"./chunk-C522H7XR.js";import"./chunk-3Y7SPD72.js";import"./chunk-ACWFKES2.js";import"./chunk-AI4VYSFS.js";import"./chunk-VJQYDYAJ.js";import"./chunk-OBAREPDC.js";import"./chunk-6GP6W46M.js";import"./chunk-PGYMKGOE.js";import"./chunk-GZLZCZ42.js";import"./chunk-GJMLK5UT.js";import"./chunk-6PX4OXQO.js";import"./chunk-AH3ZD3NJ.js";import"./chunk-HOZUDE4C.js";import"./chunk-ATHWDAXB.js";import"./chunk-N6PIIK6G.js";import"./chunk-WSF24I3D.js";import"./chunk-GHBNRMSJ.js";import"./chunk-XLZYRTNO.js";import"./chunk-EWFECZ2U.js";import"./chunk-VWNFAW2M.js";import"./chunk-XBVUVFJK.js";import"./chunk-EN2QSKFF.js";import"./chunk-OY2AHLSF.js";export{z as GNNLayerFactory,i as QEFlashAttention,x as QEGNNEmbeddingIndex,y as QEGNNIndexFactory,a as QESONA,e as QE_FLASH_ATTENTION_CONFIG,g as QE_PERFORMANCE_TARGETS,f as QE_SONA_CONFIG,A as TensorCompressionFactory,u as batchComputeAttention,K as batchDifferentiableSearch,M as checkRuvectorPackagesAvailable,c as createDomainQESONA,j as createQEFlashAttention,b as createQESONA,D as getDifferentiableSearch,F as getGetCompressionLevel,E as getHierarchicalForward,G as getInit,h as getOptimalBlockConfig,k as getQEFlashAttentionConfig,n as getRuvectorDotProductAttention,m as getRuvectorFlashAttention,p as getRuvectorHyperbolicAttention,B as getRuvectorLayer,q as getRuvectorLinearAttention,r as getRuvectorMoEAttention,o as getRuvectorMultiHeadAttention,L as getRuvectorPackageVersions,C as getTensorCompress,l as getWorkloadTypes,N as initAllRuvectorPackages,w as initGNN,d as isAttentionAvailable,v as isGNNAvailable,s as toFloat32Array,H as toFloat32ArrayGNN,J as toIEmbedding,t as toNumberArrayAttn,I as toNumberArrayGNN};

package/dist/cli/commands/fleet.js

@@ -16,6 +16,8 @@ import { createPersistentScheduler } from '../scheduler/index.js';
 import { integrateCodeIntelligence } from '../../init/fleet-integration.js';
 import { runFleetInitWizard } from '../wizards/fleet-wizard.js';
 import { FleetProgressManager, createTimedSpinner } from '../utils/progress.js';
+import { FLEET_OPERATIONS, planFleetOperation } from '../utils/fleet-operation.js';
+import { reportFleetVerdict } from '../utils/fleet-result.js';
 import { writeOutput, toJSON } from '../utils/ci-output.js';
 export function createFleetCommand(context, cleanupAndExit, ensureInitialized, registerDomainWorkflowActions) {
     const fleetCmd = new Command('fleet')
@@ -227,15 +229,18 @@ export function createFleetCommand(context, cleanupAndExit, ensureInitialized, r
             if (!Number.isSafeInteger(parallelCount) || parallelCount < 1) {
                 throw new Error('--parallel must be a positive integer');
             }
-            if (!['test', 'analyze', 'scan'].includes(operation)) {
+            if (!FLEET_OPERATIONS.includes(operation)) {
                 throw new Error(`Unsupported operation: ${operation}`);
             }
-            const operationTasks = {
-                test: ['test-generation', 'generate-tests'],
-                analyze: ['coverage-analysis', 'analyze-coverage'],
-                scan: ['security-compliance', 'scan-security'],
-            };
-            const [domain, taskType] = operationTasks[operation];
+            // #1125: each operation runs a real domain handler with the input it
+            // needs, taken from the target; without that input it is refused.
+            const plan = await planFleetOperation(operation, options.target);
+            if (!plan.success) {
+                console.error(chalk.red(`\n Operation '${operation}' cannot run: ${plan.error}`));
+                return cleanupAndExit(1);
+            }
+            const { domain, payload } = plan.task;
+            const taskType = plan.task.type;
             // #1081: refuse an operation whose task no domain plugin executes,
             // instead of submitting work that can only fail at dispatch.
             const executable = context.queen.validateTaskExecutable(taskType, [domain]);
@@ -272,7 +277,7 @@ export function createFleetCommand(context, cleanupAndExit, ensureInitialized, r
                         type: taskType,
                         priority: 'p1',
                         targetDomains: [domain],
-                        payload: { target: options.target, workerId: op.id },
+                        payload: { ...payload, target: options.target, workerId: op.id },
                         timeout: 60000,
                     });
                     if (!taskResult.success) {
@@ -287,7 +292,7 @@ export function createFleetCommand(context, cleanupAndExit, ensureInitialized, r
                             throw new Error(`Task status unavailable: ${taskResult.value}`);
                         if (task.status === 'completed') {
                             progress.completeAgent(op.id, true);
-                            return { id: op.id, success: true };
+                            return { id: op.id, success: true, result: task.result };
                         }
                         if (task.status === 'failed' || task.status === 'cancelled') {
                             throw new Error(task.error || `Task ${task.status}`);
@@ -299,7 +304,7 @@ export function createFleetCommand(context, cleanupAndExit, ensureInitialized, r
                 catch (error) {
                     console.error(chalk.red(`   ${op.name}: ${error instanceof Error ? error.message : String(error)}`));
                     progress.completeAgent(op.id, false);
-                    return { id: op.id, success: false };
+                    return { id: op.id, success: false, result: undefined };
                 }
             }));
             progress.stop();
@@ -308,12 +313,14 @@ export function createFleetCommand(context, cleanupAndExit, ensureInitialized, r
             console.log(chalk.blue('\n Operation Summary:'));
             console.log(chalk.gray(`   Operation: ${operation}`));
             console.log(chalk.gray(`   Target: ${options.target}`));
-            console.log(chalk.green(`   Successful: ${successful}`));
+            console.log(chalk.gray(`   Completed: ${successful}`));
             if (failed > 0) {
                 console.log(chalk.red(`   Failed: ${failed}`));
             }
+            // #1125 review: a completed task is not a passed check.
+            const passed = reportFleetVerdict(operation, results);
             console.log('');
-            await cleanupAndExit(failed > 0 ? 1 : 0);
+            await cleanupAndExit(passed ? 0 : 1);
         }
         catch (error) {
             console.error(chalk.red('\n Fleet operation failed:'), error);

package/dist/cli/commands/hooks-handlers/hooks-dream-learning.js

@@ -115,7 +115,8 @@ export async function checkAndTriggerDream(memoryBackend) {
                 domain: r.pattern.qeDomain || 'learning-optimization',
                 patternType: r.pattern.patternType,
                 confidence: r.pattern.confidence,
-                successRate: r.pattern.successRate || 0.5,
+                // A real 0% stays 0%; a pattern never used has no rate (#1142 review).
+                successRate: r.pattern.usageCount > 0 ? r.pattern.successRate : undefined,
             }));
             await engine.loadPatternsAsConcepts(importPatterns);
         }
@@ -746,7 +747,13 @@ export async function consolidateExperiencesToPatterns() {
                     patternId: existing.id,
                     usageCount: agg.cnt,
                     successfulUses: agg.successes,
-                    metrics: { source: 'session-consolidation', domain: agg.domain, agent: agg.agent },
+                    // #1129: these aggregates are selected by success rate (HAVING
+                    // success_rate >= 0.6) from hook-reported success, so they are uses,
+                    // never measured outcomes.
+                    metrics: {
+                        source: 'session-consolidation', domain: agg.domain, agent: agg.agent,
+                        outcome: 'unmeasured', outcomeBasis: 'session-consolidation',
+                    },
                 });
                 db.prepare(`
           UPDATE qe_patterns

package/dist/cli/commands/hooks-handlers/session-hooks.js

@@ -61,7 +61,7 @@ export function registerSessionHooks(hooks) {
             if (domainEntries.length > 0) {
                 contextParts.push(`Top domains: ${domainEntries.map(([d, c]) => `${d}(${c})`).join(', ')}`);
             }
-            if (stats.patternSuccessRate > 0) {
+            if (stats.patternSuccessRate !== null && stats.patternSuccessRate > 0) {
                 contextParts.push(`Pattern success rate: ${(stats.patternSuccessRate * 100).toFixed(0)}%`);
             }
             if (stats.routingRequests > 0) {
@@ -131,7 +131,7 @@ export function registerSessionHooks(hooks) {
             }
             if (options.json) {
                 const summary = stats
-                    ? `Session complete: ${stats.totalPatterns} patterns, ${stats.routingRequests} routings, ${(stats.patternSuccessRate * 100).toFixed(0)}% success rate`
+                    ? `Session complete: ${stats.totalPatterns} patterns, ${stats.routingRequests} routings, ${stats.patternSuccessRate === null ? 'success rate: insufficient evidence' : `${(stats.patternSuccessRate * 100).toFixed(0)}% success rate`}`
                     : 'Session complete';
                 // Stop hooks don't support hookSpecificOutput — only simple fields
                 printJson({

package/dist/cli/commands/hooks-handlers/stats-hooks.js

@@ -8,6 +8,7 @@ import chalk from 'chalk';
 import { QE_HOOK_EVENTS } from '../../../learning/qe-hooks.js';
 import { safeJsonParse } from '../../../shared/safe-json.js';
 import { committedPendingIndexPatternId } from '../../../learning/pattern-mutation-error.js';
+import { formatLearningOutcomes, formatPatternSuccessRate } from '../../../learning/usage-outcome.js';
 import { getHooksSystem, printJson, printSuccess, printError, } from './hooks-shared.js';
 /**
  * How a learned pattern reached vector search, by cause (#981, review N1).
@@ -72,8 +73,8 @@ export function registerStatsHooks(hooks) {
                 console.log(`  Requests: ${stats.routingRequests}`);
                 console.log(`  Avg Confidence: ${(stats.avgRoutingConfidence * 100).toFixed(1)}%`);
                 console.log(chalk.bold('\nLearning:'));
-                console.log(`  Outcomes: ${stats.learningOutcomes}`);
-                console.log(`  Success Rate: ${(stats.patternSuccessRate * 100).toFixed(1)}%`);
+                console.log(`  Outcomes: ${formatLearningOutcomes(stats)}`);
+                console.log(`  Success Rate: ${formatPatternSuccessRate(stats)}`);
                 console.log(chalk.bold('\nSearch Performance:'));
                 console.log(`  Operations: ${stats.patternStoreStats.searchOperations}`);
                 console.log(`  Avg Latency: ${stats.patternStoreStats.avgSearchLatencyMs.toFixed(2)}ms`);

package/dist/cli/commands/learning.js

@@ -26,6 +26,7 @@ import { createSQLitePatternStore } from '../../learning/sqlite-persistence.js';
 import { getActiveEmbeddingSpaceIdentity } from '../../learning/real-embeddings.js';
 import { inspectEmbeddingSpaceRows } from '../../learning/embedding-space.js';
 import { committedPendingIndexPatternId } from '../../learning/pattern-mutation-error.js';
+import { formatLearningOutcomes, formatPatternSuccessRate } from '../../learning/usage-outcome.js';
 // Extracted helpers
 import { initializeLearningSystem, printJson, printSuccess, printError, printInfo, displayDashboard, getDbPath, compressFile, decompressFile, verifyDatabaseIntegrity, getSchemaVersion, padRight, EXPORT_FORMAT_VERSION, DOMAIN_MAPPING, PATTERN_TYPE_MAPPING, } from './learning-helpers.js';
 // ============================================================================
@@ -241,8 +242,8 @@ function registerStatsCommand(learning) {
                 console.log(chalk.bold('\nLearning:'));
                 console.log(`  Routing Requests: ${stats.routingRequests}`);
                 console.log(`  Avg Routing Confidence: ${(stats.avgRoutingConfidence * 100).toFixed(1)}%`);
-                console.log(`  Learning Outcomes: ${stats.learningOutcomes}`);
-                console.log(`  Pattern Success Rate: ${(stats.patternSuccessRate * 100).toFixed(1)}%`);
+                console.log(`  Learning Outcomes: ${formatLearningOutcomes(stats)}`);
+                console.log(`  Pattern Success Rate: ${formatPatternSuccessRate(stats)}`);
                 console.log(chalk.bold('\nSearch Performance:'));
                 console.log(`  Operations: ${stats.patternStoreStats.searchOperations}`);
                 console.log(`  Avg Latency: ${stats.patternStoreStats.avgSearchLatencyMs.toFixed(2)}ms`);
@@ -1227,7 +1228,8 @@ function registerDreamCommand(learning) {
                     description: result.pattern.description || `${result.pattern.patternType} pattern`,
                     domain: result.pattern.qeDomain || 'learning-optimization',
                     patternType: result.pattern.patternType, confidence: result.pattern.confidence,
-                    successRate: result.pattern.successRate || 0.5,
+                    // A real 0% stays 0%; a pattern never used has no rate (#1142 review).
+                    successRate: result.pattern.usageCount > 0 ? result.pattern.successRate : undefined,
                 }));
                 loaded = await engine.loadPatternsAsConcepts(importPatterns);
                 printInfo(`Loaded ${loaded} patterns from ReasoningBank as concepts`);

package/dist/cli/commands/ruvector-commands.js

@@ -144,7 +144,8 @@ function isValidFlagName(name) {
     return VALID_FLAG_NAMES.includes(name);
 }
 function isValidProfile(name) {
-    return name in PROFILES;
+    // #1136: own entries only; `constructor` is not a profile.
+    return Object.hasOwn(PROFILES, name);
 }
 // ============================================================================
 // Command: ruvector status

package/dist/cli/commands/test.js

@@ -215,6 +215,12 @@ export function createTestCommand(context, cleanupAndExit, ensureInitialized) {
                         console.log(`    Failed: ${chalk.red(run.failed)}`);
                         console.log(`    Skipped: ${chalk.yellow(run.skipped)}`);
                         console.log(`    Duration: ${chalk.cyan(run.duration + 'ms')}`);
+                        if (run.retries) {
+                            const unmatched = run.retries.tests.filter(t => t.unmatched).length;
+                            console.log(`    Retries: ${chalk.white(run.retries.totalRetries)} run, ` +
+                                `${chalk.green(run.retries.nowPassing)} now passing, ${chalk.red(run.retries.stillFailing)} still failing` +
+                                (unmatched > 0 ? `, ${chalk.yellow(unmatched)} matched no test` : ''));
+                        }
                     }
                     // Exit codes: 1 = failures, 2 = >20% skipped, 0 = all good
                     if (run.failed > 0) {

package/dist/cli/utils/ci-output.d.ts

@@ -67,6 +67,15 @@ export interface TestRunSummary {
     skipped: number;
     duration: number;
     tests?: TestResult[];
+    /** #1130: what the retries of failed tests did, when retries were requested. */
+    retries?: {
+        totalRetries: number;
+        nowPassing: number;
+        stillFailing: number;
+        tests: Array<{
+            unmatched?: boolean;
+        }>;
+    };
 }
 export interface CoverageResult {
     /** `null` = the report never collected that metric (#995); not 0%. */

package/dist/cli/utils/fleet-operation.d.ts

@@ -0,0 +1,32 @@
+/**
+ * Agentic QE v3 - `aqe fleet run` operation tasks (#1125)
+ *
+ * Each fleet operation runs a real domain handler with the payload it
+ * requires, built from the target on disk:
+ *   - test    -> test-generation `generate-tests` with the target's source files
+ *   - scan    -> security-compliance `sast-scan` with the target's JS/TS files
+ *   - analyze -> coverage-analysis `analyze-coverage` with the coverage report
+ *                that already exists for the target
+ * When the target cannot supply that input the operation is refused with what
+ * is missing. Nothing is estimated or made up to fill a payload.
+ */
+import type { DomainName } from '../../shared/types/index.js';
+export declare const FLEET_OPERATIONS: readonly ["test", "analyze", "scan"];
+export type FleetOperation = typeof FLEET_OPERATIONS[number];
+export interface FleetOperationTask {
+    domain: DomainName;
+    type: string;
+    payload: Record<string, unknown>;
+}
+export type FleetOperationPlan = {
+    success: true;
+    task: FleetOperationTask;
+} | {
+    success: false;
+    error: string;
+};
+/** Coverage threshold `analyze` judges against, as `aqe coverage` does by default. */
+export declare const FLEET_COVERAGE_THRESHOLD = 80;
+/** Build the domain task for a fleet operation on `target`, or say why it cannot run. */
+export declare function planFleetOperation(operation: FleetOperation, target: string): Promise<FleetOperationPlan>;
+//# sourceMappingURL=fleet-operation.d.ts.map

package/dist/cli/utils/fleet-operation.js

@@ -0,0 +1,71 @@
+/**
+ * Agentic QE v3 - `aqe fleet run` operation tasks (#1125)
+ *
+ * Each fleet operation runs a real domain handler with the payload it
+ * requires, built from the target on disk:
+ *   - test    -> test-generation `generate-tests` with the target's source files
+ *   - scan    -> security-compliance `sast-scan` with the target's JS/TS files
+ *   - analyze -> coverage-analysis `analyze-coverage` with the coverage report
+ *                that already exists for the target
+ * When the target cannot supply that input the operation is refused with what
+ * is missing. Nothing is estimated or made up to fill a payload.
+ */
+import { existsSync, statSync } from 'node:fs';
+import { extname, resolve } from 'node:path';
+import { walkSourceFiles } from './file-discovery.js';
+import { summarizeCoverage } from './coverage-data.js';
+import { describeNoData, resolveExistingCoverage } from '../../coordination/handlers/coverage-target.js';
+import { MAX_SAST_FILE_BYTES } from '../../domains/security-compliance/sast-path-guard.js';
+export const FLEET_OPERATIONS = ['test', 'analyze', 'scan'];
+/** Coverage threshold `analyze` judges against, as `aqe coverage` does by default. */
+export const FLEET_COVERAGE_THRESHOLD = 80;
+/** Extensions the built-in SAST patterns cover (sast-scanner.ts). */
+const SAST_EXTENSIONS = new Set(['.ts', '.tsx', '.js', '.jsx', '.mjs', '.cjs']);
+const refuse = (error) => ({ success: false, error });
+/** Build the domain task for a fleet operation on `target`, or say why it cannot run. */
+export async function planFleetOperation(operation, target) {
+    const targetPath = resolve(target);
+    if (!existsSync(targetPath))
+        return refuse(`Target does not exist: ${target}`);
+    switch (operation) {
+        case 'test': {
+            const sourceFiles = walkSourceFiles(targetPath, { includeTests: false });
+            if (sourceFiles.length === 0)
+                return refuse(`No source files under ${target} to generate tests for.`);
+            return { success: true, task: { domain: 'test-generation', type: 'generate-tests', payload: { sourceFiles } } };
+        }
+        case 'scan': {
+            const files = walkSourceFiles(targetPath, { includeTests: true })
+                .filter(file => SAST_EXTENSIONS.has(extname(file).toLowerCase()));
+            if (files.length === 0) {
+                return refuse(`No files under ${target} that the SAST scanner covers (${[...SAST_EXTENSIONS].join(' ')}).`);
+            }
+            // sast-scan refuses the whole request when a file exceeds its limit.
+            const tooLarge = files.filter(file => statSync(file).size > MAX_SAST_FILE_BYTES);
+            if (tooLarge.length > 0) {
+                return refuse(`${tooLarge.length} files under ${target} are over the ${MAX_SAST_FILE_BYTES / (1024 * 1024)} MB ` +
+                    `SAST limit (${tooLarge.slice(0, 5).join(', ')}${tooLarge.length > 5 ? ', …' : ''}). Narrow --target to the source to scan.`);
+            }
+            return { success: true, task: { domain: 'security-compliance', type: 'sast-scan', payload: { files } } };
+        }
+        case 'analyze': {
+            const existing = await resolveExistingCoverage(targetPath);
+            if (existing.status === 'no-report') {
+                return refuse(`No coverage report for project ${existing.projectRoot}, so there is no coverage to ` +
+                    'analyze. Run your tests with coverage first (e.g. `npm test -- --coverage`), or pass the report ' +
+                    'file as --target.');
+            }
+            if (existing.status === 'target-missing')
+                return refuse(describeNoData('target-missing', target, null));
+            const { evaluation, projectRoot } = existing;
+            if (evaluation.status !== 'measured')
+                return refuse(describeNoData(evaluation.reason, target, projectRoot));
+            const coverageData = { files: evaluation.files, summary: summarizeCoverage(evaluation.files) };
+            return {
+                success: true,
+                task: { domain: 'coverage-analysis', type: 'analyze-coverage', payload: { coverageData, threshold: FLEET_COVERAGE_THRESHOLD } },
+            };
+        }
+    }
+}
+//# sourceMappingURL=fleet-operation.js.map

package/dist/cli/utils/fleet-result.d.ts

@@ -0,0 +1,31 @@
+/**
+ * Agentic QE v3 - `aqe fleet run` result summary (#1125 review)
+ *
+ * A completed task is not a passed check. The summary shows what each
+ * operation's handler returned and turns it into a verdict:
+ *   - scan    fails on a critical or high finding, or when the scan evidence
+ *             says not every requested file was analyzed;
+ *   - test    fails when no test was generated or a generated test failed
+ *             the test quality gate;
+ *   - analyze fails when coverage over the collected metrics is below the
+ *             threshold (an uncollected metric is not counted as 0%).
+ * A result without the expected shape fails rather than passing unread.
+ */
+import { type FleetOperation } from './fleet-operation.js';
+export interface FleetResultSummary {
+    lines: string[];
+    passed: boolean;
+    /** Why the verdict failed; empty when it passed. */
+    reasons: string[];
+}
+/** Summarize one completed task's result for its fleet operation. */
+export declare function summarizeFleetResult(operation: FleetOperation, result: unknown): FleetResultSummary;
+/**
+ * Print each distinct worker result once and the verdict over all workers.
+ * Returns whether the operation passed.
+ */
+export declare function reportFleetVerdict(operation: FleetOperation, workers: ReadonlyArray<{
+    success: boolean;
+    result: unknown;
+}>, log?: (line: string) => void): boolean;
+//# sourceMappingURL=fleet-result.d.ts.map

package/dist/cli/utils/fleet-result.js

@@ -0,0 +1,124 @@
+/**
+ * Agentic QE v3 - `aqe fleet run` result summary (#1125 review)
+ *
+ * A completed task is not a passed check. The summary shows what each
+ * operation's handler returned and turns it into a verdict:
+ *   - scan    fails on a critical or high finding, or when the scan evidence
+ *             says not every requested file was analyzed;
+ *   - test    fails when no test was generated or a generated test failed
+ *             the test quality gate;
+ *   - analyze fails when coverage over the collected metrics is below the
+ *             threshold (an uncollected metric is not counted as 0%).
+ * A result without the expected shape fails rather than passing unread.
+ */
+import chalk from 'chalk';
+import { FLEET_COVERAGE_THRESHOLD } from './fleet-operation.js';
+import { thresholdVerdict } from '../../coordination/handlers/coverage-target.js';
+const SEVERITIES = ['critical', 'high', 'medium', 'low', 'informational'];
+const MAX_LISTED_FINDINGS = 10;
+const asRecord = (value) => value !== null && typeof value === 'object' ? value : undefined;
+const pct = (value) => typeof value === 'number' ? `${value}%` : 'n/a';
+function summarizeScan(result) {
+    const summary = asRecord(result.summary);
+    const findings = Array.isArray(result.vulnerabilities) ? result.vulnerabilities : undefined;
+    if (!summary || !findings)
+        return unreadable('scan');
+    const count = (severity) => Number(summary[severity] ?? 0);
+    const coverage = asRecord(result.coverage);
+    const evidence = asRecord(result.evidence);
+    const completeness = typeof evidence?.completeness === 'string' ? evidence.completeness : 'unknown';
+    const lines = [
+        `Findings: ${SEVERITIES.map(s => `${s} ${count(s)}`).join(', ')}`,
+        `Files scanned: ${coverage?.filesScanned ?? 'n/a'} (evidence: ${completeness})`,
+        ...findings.slice(0, MAX_LISTED_FINDINGS).map(f => `  [${f.severity ?? '?'}] ${f.title ?? 'untitled'} @ ${f.location?.file ?? '?'}:${f.location?.line ?? '?'}`),
+    ];
+    if (findings.length > MAX_LISTED_FINDINGS)
+        lines.push(`  … ${findings.length - MAX_LISTED_FINDINGS} more`);
+    const reasons = [];
+    if (count('critical') + count('high') > 0)
+        reasons.push(`${count('critical')} critical and ${count('high')} high findings`);
+    if (completeness !== 'complete')
+        reasons.push(`scan evidence is ${completeness}: not every requested file was analyzed`);
+    return { lines, passed: reasons.length === 0, reasons };
+}
+function summarizeTests(result) {
+    if (!Array.isArray(result.tests))
+        return unreadable('test');
+    const tests = result.tests;
+    const assertions = tests.reduce((sum, t) => sum + (typeof t.assertions === 'number' ? t.assertions : 0), 0);
+    const modes = [...new Set(tests.map(t => t.generationMode).filter((m) => typeof m === 'string'))];
+    const gateFailed = tests.filter(t => asRecord(t.qualityGateResult)?.passed === false);
+    const lines = [
+        `Tests generated: ${tests.length}, assertions: ${assertions}` +
+            (modes.length > 0 ? `, generation mode: ${modes.join(', ')}` : ''),
+        `Quality gate: ${gateFailed.length === 0 ? 'passed' : `failed for ${gateFailed.length} of ${tests.length} tests`}`,
+        ...gateFailed.slice(0, MAX_LISTED_FINDINGS).map(t => `  ${t.testFile ?? t.name ?? t.id}`),
+    ];
+    const reasons = [];
+    if (tests.length === 0)
+        reasons.push('no tests were generated');
+    if (gateFailed.length > 0)
+        reasons.push(`${gateFailed.length} generated tests failed the test quality gate`);
+    return { lines, passed: reasons.length === 0, reasons };
+}
+function summarizeCoverage(result) {
+    const summary = asRecord(result.summary);
+    if (!summary || typeof summary.line !== 'number')
+        return unreadable('analyze');
+    const collected = asRecord(result.collected);
+    const isCollected = (name) => collected?.[name] !== false;
+    // An uncollected metric is 0 by construction (#1027 B2): show it as n/a and
+    // leave it out of the verdict, as `aqe coverage` does (#995).
+    const metric = (name) => isCollected(name) ? pct(summary[name]) : 'n/a (not collected)';
+    const num = (name) => typeof summary[name] === 'number' ? summary[name] : 0;
+    const verdict = thresholdVerdict({ line: num('line'), branch: num('branch'), function: num('function'), statement: num('statement') }, FLEET_COVERAGE_THRESHOLD, { branch: isCollected('branch'), function: isCollected('function'), statement: isCollected('statement') });
+    const lines = [
+        `Coverage: line ${pct(summary.line)}, branch ${metric('branch')}, function ${metric('function')}, ` +
+            `statement ${metric('statement')} over ${summary.files ?? 'n/a'} files`,
+        `Overall over collected metrics: ${verdict.overall.toFixed(1)}% (threshold ${FLEET_COVERAGE_THRESHOLD}%)`,
+    ];
+    const reasons = verdict.meetsThreshold ? [] : [`coverage is below the ${FLEET_COVERAGE_THRESHOLD}% threshold`];
+    return { lines, passed: reasons.length === 0, reasons };
+}
+function unreadable(operation) {
+    const reason = `the ${operation} result has no readable summary`;
+    return { lines: [`Result: ${reason}`], passed: false, reasons: [reason] };
+}
+/** Summarize one completed task's result for its fleet operation. */
+export function summarizeFleetResult(operation, result) {
+    const record = asRecord(result);
+    if (!record)
+        return unreadable(operation);
+    switch (operation) {
+        case 'scan': return summarizeScan(record);
+        case 'test': return summarizeTests(record);
+        case 'analyze': return summarizeCoverage(record);
+    }
+}
+/**
+ * Print each distinct worker result once and the verdict over all workers.
+ * Returns whether the operation passed.
+ */
+export function reportFleetVerdict(operation, workers, log = line => console.log(line)) {
+    const summaries = workers.filter(w => w.success).map(w => summarizeFleetResult(operation, w.result));
+    const shown = new Set();
+    for (const summary of summaries) {
+        const text = summary.lines.join('\n');
+        if (shown.has(text))
+            continue;
+        shown.add(text);
+        log('');
+        for (const line of summary.lines)
+            log(`   ${line}`);
+    }
+    const incomplete = workers.filter(w => !w.success).length;
+    const reasons = [...new Set(summaries.flatMap(s => s.reasons))];
+    if (incomplete > 0)
+        reasons.unshift(`${incomplete} of ${workers.length} workers did not complete`);
+    log('');
+    log(reasons.length === 0
+        ? chalk.green('   Verdict: PASSED')
+        : chalk.red(`   Verdict: FAILED (${reasons.join('; ')})`));
+    return reasons.length === 0;
+}
+//# sourceMappingURL=fleet-result.js.map

package/dist/context/sources/git-source.js

@@ -1,7 +1,7 @@
 /**
  * Git Source - Recent commit context for target files
  */
-import { execFileSync } from 'node:child_process';
+import { runGit } from '../../shared/git/git-process.js';
 export class GitContextSource {
     cwd;
     id = 'git';
@@ -18,7 +18,11 @@ export class GitContextSource {
                 // filePaths can originate from MCP tool arguments. Pass each path as
                 // one argv element; interpolating it into a shell command would allow
                 // command execution via quotes or shell metacharacters.
-                const log = execFileSync('git', ['log', '--oneline', '-5', '--', file], { cwd: this.cwd ?? process.cwd(), encoding: 'utf-8', timeout: 5000, stdio: ['ignore', 'pipe', 'ignore'] }).trim();
+                // Async, in its own process group, so a slow git can neither stall
+                // the MCP server's event loop nor outlive its timeout (#1115).
+                const log = (await runGit(['log', '--oneline', '-5', '--', file], {
+                    cwd: this.cwd ?? process.cwd(), timeout: 5000,
+                })).trim();
                 if (log) {
                     const content = `Recent commits for ${file}:\n${log}`;
                     fragments.push({