agentic-qe 3.9.14 → 3.9.15

package/dist/validation/command-eval-runner.d.ts

@@ -0,0 +1,154 @@
+/**
+ * Command Eval Runner — shell-command-based eval suite execution.
+ *
+ * Sibling to {@link ParallelEvalRunner} (which evaluates LLM prompts against
+ * keyword-match patterns). This runner is for skills whose eval suite is a
+ * set of shell commands that produce JSON envelopes — e.g. qe-browser, whose
+ * five primitives (assert, batch, visual-diff, check-injection, intent-score)
+ * are Node scripts that exit 0/1 with structured stdout.
+ *
+ * Detection is by suite shape, not by an explicit type tag: if the first test
+ * case has `input.command`, we treat the suite as command-mode.
+ *
+ * Supported assertions per test case:
+ *   - `exit_code`: strict equality against the process exit code
+ *   - `json_fields`: a map of dotted JSONPath -> expected value, evaluated
+ *     against parsed stdout JSON
+ *   - `severity_at_least`: ordered comparison against
+ *     `.output.checkInjection.severity` in the range
+ *     `none < low < medium < high < critical`
+ *   - `candidate_count_at_least`: numeric lower bound on
+ *     `.output.intentScore.candidateCount`
+ *
+ * Setup steps in `input.setup[]` run sequentially before the main command;
+ * any non-zero setup exit short-circuits the test case as failed.
+ */
+declare const SEVERITY_ORDER: readonly ["none", "low", "medium", "high", "critical"];
+export type Severity = (typeof SEVERITY_ORDER)[number];
+/**
+ * Return true iff `actual` is at or above `threshold` on the severity scale.
+ * Unknown severities return false (conservative — we never silently pass).
+ */
+export declare function severityAtLeast(actual: unknown, threshold: Severity): boolean;
+/**
+ * Evaluate a simple dotted JSONPath like `.output.assert.passed` against an
+ * already-parsed JSON value. Returns `undefined` if any segment is missing.
+ *
+ * We intentionally support only dotted paths (no `[idx]`, no filters) because
+ * that's the full surface area used by qe-browser's output schemas.
+ */
+export declare function evalJsonPath(obj: unknown, jsonPath: string): unknown;
+export type TestCasePriority = 'critical' | 'high' | 'medium' | 'low';
+export interface CommandEvalExpected {
+    exit_code?: number;
+    /** Dotted JSONPath → expected JSON value (deep equality). */
+    json_fields?: Record<string, unknown>;
+    severity_at_least?: Severity;
+    candidate_count_at_least?: number;
+}
+export interface CommandEvalTestCase {
+    id: string;
+    description?: string;
+    category?: string;
+    priority?: TestCasePriority;
+    input: {
+        setup?: string[];
+        command: string;
+    };
+    expected: CommandEvalExpected;
+}
+export interface CommandEvalSuite {
+    skill: string;
+    version?: string;
+    status?: 'design-spec' | 'active' | 'deprecated';
+    description?: string;
+    test_cases: CommandEvalTestCase[];
+    validation?: {
+        required_pass_rate?: number;
+        critical_must_pass?: boolean;
+    };
+}
+export interface CommandEvalTestResult {
+    testId: string;
+    passed: boolean;
+    category?: string;
+    priority?: TestCasePriority;
+    exitCode: number | null;
+    stdoutSnippet: string;
+    stderrSnippet: string;
+    /** Assertion failures — empty when passed. */
+    failures: string[];
+    durationMs: number;
+    /** Set when a setup step failed before the main command ran. */
+    setupFailure?: string;
+}
+export interface CommandEvalResult {
+    skill: string;
+    passed: boolean;
+    totalTests: number;
+    passedTests: number;
+    failedTests: number;
+    passRate: number;
+    /** True iff every test with priority=critical passed. */
+    criticalPassed: boolean;
+    testResults: CommandEvalTestResult[];
+    totalDurationMs: number;
+    timestamp: Date;
+}
+/** Injectable runner so tests don't spawn real processes. */
+export type CommandRunner = (cmd: string, options: {
+    timeoutMs: number;
+    cwd: string;
+}) => {
+    status: number | null;
+    stdout: string;
+    stderr: string;
+    error?: Error;
+};
+/**
+ * A suite is command-eval mode when the first test case has `input.command`.
+ * We use this to dispatch from the shared `aqe eval` CLI without requiring a
+ * new explicit type tag in every existing LLM-style yaml.
+ */
+export declare function isCommandEvalSuite(raw: unknown): raw is CommandEvalSuite;
+/**
+ * Validate a single test-case result against its expected block.
+ * Returns the list of failures (empty == passed).
+ *
+ * Parses stdout as JSON only if at least one assertion needs it, so commands
+ * that only care about `exit_code` don't require JSON output.
+ */
+export declare function validateCommandResult(tc: CommandEvalTestCase, exitCode: number | null, stdout: string): {
+    passed: boolean;
+    failures: string[];
+};
+export interface CommandEvalRunnerOptions {
+    /** Directory containing skill directories (default: `.claude/skills`). */
+    skillsDir?: string;
+    /** Per-command timeout in milliseconds (default: 60s). */
+    timeoutMs?: number;
+    /** Working directory for commands (default: `process.cwd()`). */
+    cwd?: string;
+    /** Injected runner for tests. */
+    runner?: CommandRunner;
+}
+export declare class CommandEvalRunner {
+    private readonly skillsDir;
+    private readonly timeoutMs;
+    private readonly cwd;
+    private readonly runner;
+    constructor(options?: CommandEvalRunnerOptions);
+    /**
+     * Load `${skill}/evals/${skill}.yaml` and return it iff it's a command-mode
+     * suite. Returns null if missing, unparseable, or not command-mode.
+     */
+    loadSuite(skill: string): CommandEvalSuite | null;
+    run(skill: string): Promise<CommandEvalResult>;
+    private executeCase;
+}
+/**
+ * Convenience factory.
+ */
+export declare function createCommandEvalRunner(options?: CommandEvalRunnerOptions): CommandEvalRunner;
+export {};
+//# sourceMappingURL=command-eval-runner.d.ts.map

package/dist/validation/command-eval-runner.js

@@ -0,0 +1,293 @@
+/**
+ * Command Eval Runner — shell-command-based eval suite execution.
+ *
+ * Sibling to {@link ParallelEvalRunner} (which evaluates LLM prompts against
+ * keyword-match patterns). This runner is for skills whose eval suite is a
+ * set of shell commands that produce JSON envelopes — e.g. qe-browser, whose
+ * five primitives (assert, batch, visual-diff, check-injection, intent-score)
+ * are Node scripts that exit 0/1 with structured stdout.
+ *
+ * Detection is by suite shape, not by an explicit type tag: if the first test
+ * case has `input.command`, we treat the suite as command-mode.
+ *
+ * Supported assertions per test case:
+ *   - `exit_code`: strict equality against the process exit code
+ *   - `json_fields`: a map of dotted JSONPath -> expected value, evaluated
+ *     against parsed stdout JSON
+ *   - `severity_at_least`: ordered comparison against
+ *     `.output.checkInjection.severity` in the range
+ *     `none < low < medium < high < critical`
+ *   - `candidate_count_at_least`: numeric lower bound on
+ *     `.output.intentScore.candidateCount`
+ *
+ * Setup steps in `input.setup[]` run sequentially before the main command;
+ * any non-zero setup exit short-circuits the test case as failed.
+ */
+import * as fs from 'node:fs';
+import * as path from 'node:path';
+import * as yaml from 'yaml';
+import { spawnSync as realSpawnSync } from 'node:child_process';
+import { toErrorMessage } from '../shared/error-utils.js';
+// ============================================================================
+// Severity ordering
+// ============================================================================
+const SEVERITY_ORDER = ['none', 'low', 'medium', 'high', 'critical'];
+/**
+ * Return true iff `actual` is at or above `threshold` on the severity scale.
+ * Unknown severities return false (conservative — we never silently pass).
+ */
+export function severityAtLeast(actual, threshold) {
+    if (typeof actual !== 'string')
+        return false;
+    const a = SEVERITY_ORDER.indexOf(actual);
+    const t = SEVERITY_ORDER.indexOf(threshold);
+    return a >= 0 && t >= 0 && a >= t;
+}
+// ============================================================================
+// JSON path evaluation
+// ============================================================================
+/**
+ * Evaluate a simple dotted JSONPath like `.output.assert.passed` against an
+ * already-parsed JSON value. Returns `undefined` if any segment is missing.
+ *
+ * We intentionally support only dotted paths (no `[idx]`, no filters) because
+ * that's the full surface area used by qe-browser's output schemas.
+ */
+export function evalJsonPath(obj, jsonPath) {
+    if (typeof jsonPath !== 'string' || !jsonPath.startsWith('.'))
+        return undefined;
+    const parts = jsonPath === '.' ? [] : jsonPath.slice(1).split('.');
+    let cur = obj;
+    for (const part of parts) {
+        if (cur == null || typeof cur !== 'object')
+            return undefined;
+        cur = cur[part];
+    }
+    return cur;
+}
+const defaultCommandRunner = (cmd, opts) => {
+    const result = realSpawnSync('bash', ['-c', cmd], {
+        timeout: opts.timeoutMs,
+        cwd: opts.cwd,
+        encoding: 'utf8',
+        maxBuffer: 16 * 1024 * 1024,
+    });
+    return {
+        status: result.status,
+        stdout: typeof result.stdout === 'string' ? result.stdout : '',
+        stderr: typeof result.stderr === 'string' ? result.stderr : '',
+        error: result.error,
+    };
+};
+// ============================================================================
+// Suite detection
+// ============================================================================
+/**
+ * A suite is command-eval mode when the first test case has `input.command`.
+ * We use this to dispatch from the shared `aqe eval` CLI without requiring a
+ * new explicit type tag in every existing LLM-style yaml.
+ */
+export function isCommandEvalSuite(raw) {
+    if (!raw || typeof raw !== 'object')
+        return false;
+    const obj = raw;
+    if (!Array.isArray(obj.test_cases) || obj.test_cases.length === 0)
+        return false;
+    const first = obj.test_cases[0];
+    return typeof first?.input?.command === 'string';
+}
+// ============================================================================
+// Result validation
+// ============================================================================
+/**
+ * Validate a single test-case result against its expected block.
+ * Returns the list of failures (empty == passed).
+ *
+ * Parses stdout as JSON only if at least one assertion needs it, so commands
+ * that only care about `exit_code` don't require JSON output.
+ */
+export function validateCommandResult(tc, exitCode, stdout) {
+    const failures = [];
+    const expected = tc.expected || {};
+    if (expected.exit_code !== undefined && exitCode !== expected.exit_code) {
+        failures.push(`exit_code: expected ${expected.exit_code}, got ${exitCode}`);
+    }
+    const needsJson = !!expected.json_fields ||
+        expected.severity_at_least !== undefined ||
+        expected.candidate_count_at_least !== undefined;
+    let parsed = null;
+    if (needsJson) {
+        try {
+            parsed = JSON.parse(stdout);
+        }
+        catch (e) {
+            failures.push(`stdout is not valid JSON: ${toErrorMessage(e)}`);
+            return { passed: false, failures };
+        }
+    }
+    if (expected.json_fields) {
+        for (const [jsonPath, want] of Object.entries(expected.json_fields)) {
+            const got = evalJsonPath(parsed, jsonPath);
+            if (!deepEqual(got, want)) {
+                failures.push(`json_fields ${jsonPath}: expected ${JSON.stringify(want)}, got ${JSON.stringify(got)}`);
+            }
+        }
+    }
+    if (expected.severity_at_least !== undefined) {
+        const sev = evalJsonPath(parsed, '.output.checkInjection.severity');
+        if (!severityAtLeast(sev, expected.severity_at_least)) {
+            failures.push(`severity_at_least: expected >= ${expected.severity_at_least}, got ${JSON.stringify(sev)}`);
+        }
+    }
+    if (expected.candidate_count_at_least !== undefined) {
+        const count = evalJsonPath(parsed, '.output.intentScore.candidateCount');
+        const countNum = typeof count === 'number' ? count : Number.NaN;
+        if (!(countNum >= expected.candidate_count_at_least)) {
+            failures.push(`candidate_count_at_least: expected >= ${expected.candidate_count_at_least}, got ${JSON.stringify(count)}`);
+        }
+    }
+    return { passed: failures.length === 0, failures };
+}
+/**
+ * Strict structural equality for JSON values — used to compare
+ * `json_fields` expected values against extracted paths. Avoids the pitfalls
+ * of `JSON.stringify` comparison (e.g. key order variance on objects) by
+ * walking both sides.
+ */
+function deepEqual(a, b) {
+    if (a === b)
+        return true;
+    if (a == null || b == null)
+        return a === b;
+    if (typeof a !== typeof b)
+        return false;
+    if (Array.isArray(a)) {
+        if (!Array.isArray(b) || a.length !== b.length)
+            return false;
+        return a.every((v, i) => deepEqual(v, b[i]));
+    }
+    if (typeof a === 'object') {
+        if (typeof b !== 'object' || Array.isArray(b))
+            return false;
+        const ao = a;
+        const bo = b;
+        const ak = Object.keys(ao);
+        const bk = Object.keys(bo);
+        if (ak.length !== bk.length)
+            return false;
+        return ak.every((k) => deepEqual(ao[k], bo[k]));
+    }
+    return false;
+}
+export class CommandEvalRunner {
+    skillsDir;
+    timeoutMs;
+    cwd;
+    runner;
+    constructor(options = {}) {
+        this.skillsDir = options.skillsDir ?? '.claude/skills';
+        this.timeoutMs = options.timeoutMs ?? 60_000;
+        this.cwd = options.cwd ?? process.cwd();
+        this.runner = options.runner ?? defaultCommandRunner;
+    }
+    /**
+     * Load `${skill}/evals/${skill}.yaml` and return it iff it's a command-mode
+     * suite. Returns null if missing, unparseable, or not command-mode.
+     */
+    loadSuite(skill) {
+        const baseDir = path.isAbsolute(this.skillsDir)
+            ? this.skillsDir
+            : path.join(this.cwd, this.skillsDir);
+        const evalPath = path.join(baseDir, skill, 'evals', `${skill}.yaml`);
+        if (!fs.existsSync(evalPath))
+            return null;
+        try {
+            const raw = yaml.parse(fs.readFileSync(evalPath, 'utf-8'));
+            if (!isCommandEvalSuite(raw))
+                return null;
+            return raw;
+        }
+        catch {
+            return null;
+        }
+    }
+    async run(skill) {
+        const start = Date.now();
+        const suite = this.loadSuite(skill);
+        if (!suite) {
+            throw new Error(`Command-eval suite not found or not command-mode at ${this.skillsDir}/${skill}/evals/${skill}.yaml`);
+        }
+        const results = [];
+        // Sequential by design: setup steps (e.g. `vibium go <url>`) mutate
+        // shared browser state, so interleaving test cases would be racy.
+        for (const tc of suite.test_cases) {
+            results.push(this.executeCase(tc));
+        }
+        const passedTests = results.filter((r) => r.passed).length;
+        const failedTests = results.length - passedTests;
+        const passRate = results.length > 0 ? passedTests / results.length : 0;
+        const requiredPassRate = suite.validation?.required_pass_rate ?? 0.9;
+        const criticalMustPass = suite.validation?.critical_must_pass ?? true;
+        const criticalResults = results.filter((r) => r.priority === 'critical');
+        const criticalPassed = criticalResults.every((r) => r.passed);
+        const passed = passRate >= requiredPassRate && (!criticalMustPass || criticalPassed);
+        return {
+            skill,
+            passed,
+            totalTests: results.length,
+            passedTests,
+            failedTests,
+            passRate,
+            criticalPassed,
+            testResults: results,
+            totalDurationMs: Date.now() - start,
+            timestamp: new Date(),
+        };
+    }
+    executeCase(tc) {
+        const start = Date.now();
+        const setup = tc.input?.setup ?? [];
+        for (const step of setup) {
+            const r = this.runner(step, { timeoutMs: this.timeoutMs, cwd: this.cwd });
+            if (r.status !== 0) {
+                return {
+                    testId: tc.id,
+                    passed: false,
+                    category: tc.category,
+                    priority: tc.priority,
+                    exitCode: r.status,
+                    stdoutSnippet: truncate(r.stdout),
+                    stderrSnippet: truncate(r.stderr),
+                    failures: [],
+                    setupFailure: `setup step failed (exit ${r.status}): ${step}`,
+                    durationMs: Date.now() - start,
+                };
+            }
+        }
+        const r = this.runner(tc.input.command, { timeoutMs: this.timeoutMs, cwd: this.cwd });
+        const v = validateCommandResult(tc, r.status, r.stdout);
+        return {
+            testId: tc.id,
+            passed: v.passed,
+            category: tc.category,
+            priority: tc.priority,
+            exitCode: r.status,
+            stdoutSnippet: truncate(r.stdout),
+            stderrSnippet: truncate(r.stderr),
+            failures: v.failures,
+            durationMs: Date.now() - start,
+        };
+    }
+}
+function truncate(s, max = 500) {
+    if (s.length <= max)
+        return s;
+    return `${s.slice(0, max)}…[truncated ${s.length - max}b]`;
+}
+/**
+ * Convenience factory.
+ */
+export function createCommandEvalRunner(options = {}) {
+    return new CommandEvalRunner(options);
+}
+//# sourceMappingURL=command-eval-runner.js.map

package/dist/validation/index.d.ts

@@ -14,6 +14,8 @@ export { SwarmSkillValidator, createSwarmSkillValidator, DEFAULT_SWARM_VALIDATIO
 export type { SwarmValidationConfig, SwarmTopology, SwarmValidationResult, SwarmValidationSummary, SkillValidatorFn, } from './swarm-skill-validator.js';
 export { ParallelEvalRunner, createParallelEvalRunner, DEFAULT_PARALLEL_EVAL_CONFIG, MockLLMExecutor, } from './parallel-eval-runner.js';
 export type { ParallelEvalConfig, EvalTestCase, EvalTestCaseInput, EvalTestCaseExpectedOutput, EvalTestCaseValidation, EvalSuite, TestCaseTask, ParallelEvalResult, WorkerMessage, WorkerProgress, EvalProgress, LLMExecutor, } from './parallel-eval-runner.js';
+export { CommandEvalRunner, createCommandEvalRunner, evalJsonPath, severityAtLeast, isCommandEvalSuite, validateCommandResult, } from './command-eval-runner.js';
+export type { CommandEvalSuite, CommandEvalTestCase, CommandEvalExpected, CommandEvalResult, CommandEvalTestResult, CommandEvalRunnerOptions, CommandRunner, Severity, TestCasePriority, } from './command-eval-runner.js';
 export { ValidationResultAggregator, createValidationResultAggregator, } from './validation-result-aggregator.js';
 export type { AggregatedValidationReport, AggregatorConfig, CrossModelReport, IssueSeverity, ModelAnomaly, ParallelValidationRunResult, RegressionReport, SkillValidationSummary, ValidationIssue, } from './validation-result-aggregator.js';
 export { TriggerOptimizer, createTriggerOptimizer, parseSkillFrontmatter, } from './trigger-optimizer.js';

package/dist/validation/index.js

@@ -25,6 +25,10 @@ P0_SKILLS, DEFAULT_VALIDATION_MODELS, } from './swarm-skill-validator.js';
 // ============================================================================
 export { ParallelEvalRunner, createParallelEvalRunner, DEFAULT_PARALLEL_EVAL_CONFIG, MockLLMExecutor, } from './parallel-eval-runner.js';
 // ============================================================================
+// Command Eval Runner (ADR-091: qe-browser and other shell-based skills)
+// ============================================================================
+export { CommandEvalRunner, createCommandEvalRunner, evalJsonPath, severityAtLeast, isCommandEvalSuite, validateCommandResult, } from './command-eval-runner.js';
+// ============================================================================
 // Validation Result Aggregator (ADR-056 Phase 5)
 // ============================================================================
 export { ValidationResultAggregator, createValidationResultAggregator, } from './validation-result-aggregator.js';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentic-qe",
-  "version": "3.9.14",
+  "version": "3.9.15",
   "description": "Agentic Quality Engineering V3 - Domain-Driven Design Architecture with 13 Bounded Contexts, O(log n) coverage analysis, ReasoningBank learning, 60 specialized QE agents, mathematical Coherence verification, deep Claude Flow integration",
   "type": "module",
   "main": "./dist/index.js",

package/dist/cli/chunks/adapter-W2EM7ZQ7.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c}from"./chunk-5U6M4XCP.js";import"./chunk-AEQMQVDJ.js";import"./chunk-QUK34CSU.js";import"./chunk-DMZ2XOEW.js";import"./chunk-UIYK4Y7K.js";import"./chunk-DGUAAG3T.js";export{a as AgentBoosterAdapter,b as createAgentBoosterAdapter,c as createAgentBoosterAdapterSync};

package/dist/cli/chunks/aqe-learning-engine-JQMPNHQU.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c,d,e,f}from"./chunk-JOGKQTCU.js";import"./chunk-7732TLNA.js";import"./chunk-3PZ2QLLR.js";import"./chunk-NWZYULZE.js";import"./chunk-KJB6JJXT.js";import"./chunk-S2KGZJHS.js";import"./chunk-NWGIQCCO.js";import"./chunk-UAAJ3BSW.js";import"./chunk-ZCRLZINE.js";import"./chunk-TJOZPPZR.js";import"./chunk-4EQWK6P6.js";import"./chunk-Y3X5OS5F.js";import"./chunk-WSZVSDD5.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{b as AQELearningEngine,a as DEFAULT_ENGINE_CONFIG,c as DEFAULT_META_LEARNING_CONFIG,d as MetaLearningEngine,e as createAQELearningEngine,f as createDefaultLearningEngine};

package/dist/cli/chunks/base-EOPYE6JR.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c,d,e,f}from"./chunk-BNPJMPM3.js";import"./chunk-VQB5SLXZ.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{f as MCPToolBase,e as defaultToolLogger,d as getMemoryBackend,b as getSharedMemoryBackend,a as registerRvfResetFn,c as resetSharedMemoryBackend};

package/dist/cli/chunks/browser-workflow-CSARLAJD.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a}from"./chunk-FFACJODG.js";import"./chunk-BNPJMPM3.js";import"./chunk-VQB5SLXZ.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{a as BrowserWorkflowTool};

package/dist/cli/chunks/chunk-53KLVMAT.js

@@ -1,18 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a as x,c as N}from"./chunk-UIYK4Y7K.js";import{b as w,c as P,d as O}from"./chunk-JNLKLC7G.js";N();O();import*as k from"fs";import*as T from"path";import*as L from"yaml";var D=5,F=4,U=3e4,W=".claude/skills",G=5e3,$=500,A=100,V=1e3,I=500,H=.8,Q=.2,K=.1,v=.1,q=.1,B=.1,z=.1,X=200,j=500,Y=300,J=1,Z=.9,M={maxWorkers:D,batchSize:F,retryFailedTests:!0,timeout:U,skillsDir:W,progressIntervalMs:G},R=class{async execute(e,s,t){let r=P(A,$+A);return await new Promise(i=>setTimeout(i,r)),{output:this.generateMockResponse(e),tokensUsed:w(I,V+I),durationMs:r}}generateMockResponse(e){let s=[];return e.includes("alt")&&s.push("alt","1.1.1","perceivable"),e.includes("contrast")&&s.push("contrast","1.4.3","4.5:1"),e.includes("keyboard")&&s.push("keyboard","2.1.1","operable","button"),e.includes("label")&&s.push("label","3.3.2","understandable"),e.includes("ARIA")&&s.push("ARIA","4.1.2","robust"),e.includes("focus")&&s.push("focus","2.4.7","outline"),e.includes("lang")&&s.push("lang","3.1.1","language"),e.includes("caption")&&s.push("caption","1.2.2","track"),e.includes("accessible")&&s.push("accessible","compliant"),s.length===0&&s.push("accessibility","WCAG","finding"),`Analysis complete. Found issues related to: ${s.join(", ")}.
-    Recommendations: Implement proper ${s[0]} attributes for better accessibility.
-    Severity: ${e.includes("critical")?"critical":"serious"}`}},y=class{constructor(e,s,t){this.executor=s;this.config=t;this.id=e}executor;config;id;status="idle";tasksCompleted=0;tasksFailed=0;lastHeartbeat=Date.now();async executeBatch(e){this.status="running";let s=[];for(let t of e){this.lastHeartbeat=Date.now();try{let r=await this.executeTestCase(t);if(s.push(r),r.passed)this.tasksCompleted++;else if(this.tasksFailed++,this.config.retryFailedTests&&!r.passed){let a=await this.executeTestCase(t,!0);a.passed&&(s[s.length-1]=a,this.tasksCompleted++,this.tasksFailed--)}}catch(r){this.tasksFailed++,s.push({testId:t.testCaseId,passed:!1,expectedPatterns:t.testCase.expected_output.must_contain||[],actualPatterns:[],reasoningQuality:0,category:t.testCase.category,priority:t.testCase.priority,error:x(r)})}}return this.status="idle",s}async executeTestCase(e,s=!1){let t=Date.now(),r=this.buildPrompt(e),a=await this.executor.execute(r,e.model,{timeout:this.config.timeout}),i=this.validateResponse(a.output,e.testCase.expected_output,e.testCase.validation);return{testId:e.testCaseId,passed:i.passed,expectedPatterns:e.testCase.expected_output.must_contain||[],actualPatterns:i.foundPatterns,reasoningQuality:i.reasoningQuality,executionTimeMs:Date.now()-t,category:e.testCase.category,priority:e.testCase.priority,error:i.error}}buildPrompt(e){let{testCase:s}=e,t=s.input.context;return`Analyze the following ${t.language||"HTML"} code for accessibility issues.
-Context: ${t.description||s.description}
-WCAG Level: ${t.wcagLevel||"AA"}
-
-Code:
-\`\`\`${t.language||"html"}
-${s.input.code}
-\`\`\`
-
-Provide a detailed analysis including:
-1. All accessibility issues found
-2. WCAG success criteria violated
-3. Severity classification (critical/serious/moderate/minor)
-4. Remediation recommendations with code examples`}validateResponse(e,s,t){let r=[],a=[],i=s.must_contain||[],o=0;for(let u of i)new RegExp(u,"i").test(e)&&(r.push(u),o++);let n=s.must_not_contain||[];for(let u of n)new RegExp(u,"i").test(e)&&a.push(`Output should not contain: ${u}`);let d=s.must_match_regex||[];for(let u of d)new RegExp(u,"i").test(e)||a.push(`Output should match regex: ${u}`);let p=i.length>0?o/i.length:1,g=t?.keyword_match_threshold??H,h=p>=g,l=this.calculateReasoningQuality(e,t),m=t?.reasoning_quality_min??0,_=l>=m;return{passed:h&&_&&a.length===0,foundPatterns:r,reasoningQuality:l,error:a.length>0?a.join("; "):void 0}}calculateReasoningQuality(e,s){let t=0;if(e.length>X&&(t+=Q),e.length>j&&(t+=K),e.includes(`
-`)&&(t+=v),(e.includes("1.")||e.includes("-"))&&(t+=v),e.includes("```")&&(t+=v),e.includes("WCAG")&&(t+=q),(e.includes("remediation")||e.includes("recommendation"))&&(t+=B),(e.includes("severity")||e.includes("critical"))&&(t+=z),s?.grading_rubric){let r=s.grading_rubric,a=0;r.completeness&&e.length>Y&&(a+=r.completeness),r.accuracy&&e.includes("WCAG")&&(a+=r.accuracy),r.actionability&&(e.includes("fix")||e.includes("change"))&&(a+=r.actionability),t=Math.max(t,a)}return Math.min(t,1)}getProgress(e){return{workerId:this.id,tasksCompleted:this.tasksCompleted,tasksTotal:e,elapsedMs:Date.now()-this.lastHeartbeat}}},S=class{constructor(e=M,s,t=new R){this.config=e;this.skillValidationLearner=s;this.executor=t;for(let r=0;r<e.maxWorkers;r++)this.workers.push(new y(r,t,e))}config;skillValidationLearner;executor;workers=[];progressCallback;onProgress(e){this.progressCallback=e}async runEvalParallel(e,s){let t=Date.now(),r=this.loadEvalSuite(e);if(!r)throw new Error(`Failed to load eval suite for skill: ${e}`);let a=this.createTasks(r,s),i=this.partitionTestCases(a,this.config.batchSize),o=a.length*(this.config.timeout/2),n=[],d=[];for(let c=0;c<i.length;c++){let b=this.workers[c%this.workers.length];d.push(b.executeBatch(i[c]))}let p=this.startProgressReporting(e,s,a.length,t);try{let c=await Promise.all(d);for(let b of c)n.push(...b)}finally{p&&clearInterval(p)}let g=Date.now()-t,h=n.filter(c=>c.passed).length,l=n.filter(c=>!c.passed).length,m=a.length-n.length,_=n.length>0?h/n.length:0,C=n.length>0?n.reduce((c,b)=>c+b.reasoningQuality,0)/n.length:0,u=g>0?o/g:1,f=_>=r.success_criteria.pass_rate&&C>=(r.success_criteria.avg_reasoning_quality??0);return await this.recordOutcome({skillName:e,trustTier:this.determineTrustTier(r),validationLevel:"eval",model:s,passed:f,score:_,testCaseResults:n,timestamp:new Date,runId:`${e}-${s}-${Date.now()}`,metadata:{duration:g,parallelSpeedup:u,workersUsed:this.config.maxWorkers,version:r.version}}),{skill:e,model:s,totalTests:a.length,passedTests:h,failedTests:l,skippedTests:m,passRate:_,testResults:n,totalDurationMs:g,parallelSpeedup:u,avgReasoningQuality:C,passed:f,workersUsed:this.config.maxWorkers,timestamp:new Date}}async runMultipleEvalsParallel(e,s){let t=new Map,r=[];for(let o of e){t.set(o,[]);for(let n of s)r.push({skill:o,model:n})}let a=r.map(async({skill:o,model:n})=>{let d=await this.runEvalParallel(o,n);return{skill:o,result:d}}),i=await Promise.all(a);for(let{skill:o,result:n}of i)t.get(o).push(n);return t}loadEvalSuite(e){let s=T.isAbsolute(this.config.skillsDir)?this.config.skillsDir:T.join(process.cwd(),this.config.skillsDir),t=T.join(s,e,"evals",`${e}.yaml`);if(!k.existsSync(t))return console.error(`Eval suite not found: ${t}`),null;try{let r=k.readFileSync(t,"utf-8");return L.parse(r)}catch(r){return console.error(`Failed to parse eval suite: ${r instanceof Error?r.message:r}`),null}}createTasks(e,s){return e.test_cases.map((t,r)=>({skillName:e.skill,testCaseId:t.id,testCase:t,model:s,batchId:Math.floor(r/this.config.batchSize),indexInBatch:r%this.config.batchSize}))}partitionTestCases(e,s){let t=[];for(let r=0;r<e.length;r+=s)t.push(e.slice(r,r+s));return t}determineTrustTier(e){let s=e.mcp_integration?.enabled??!1,t=e.learning?.cross_model_comparison??!1,r=e.success_criteria.critical_pass_rate??0;return s&&t&&r>=J?3:s||r>=Z?2:1}startProgressReporting(e,s,t,r){return this.progressCallback?setInterval(()=>{let a=Date.now()-r,i=this.workers.reduce((l,m)=>l+m.tasksCompleted,0),o=this.workers.reduce((l,m)=>l+m.tasksFailed,0),n=this.workers.filter(l=>l.status==="running").length,d=a>0?i/a:0,p=t-i,g=d>0?p/d:0,h={skill:e,model:s,totalTasks:t,completedTasks:i,failedTasks:o,activeWorkers:n,elapsedMs:a,estimatedRemainingMs:g,workerProgress:this.workers.map(l=>l.getProgress(t))};this.progressCallback(h)},this.config.progressIntervalMs):null}async recordOutcome(e){await this.skillValidationLearner.recordValidationOutcome(e)}getWorkerStatus(){return this.workers.map(e=>({id:e.id,status:e.status,tasksCompleted:e.tasksCompleted,tasksFailed:e.tasksFailed}))}resetWorkers(){for(let e of this.workers)e.tasksCompleted=0,e.tasksFailed=0,e.status="idle"}};function re(E,e={},s){let t={...M,...e};return new S(t,E,s)}export{M as a,re as b};

package/dist/cli/chunks/chunk-STYNOKOM.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a as e,b as g,e as r}from"./chunk-GUEK5DOK.js";import{c as o}from"./chunk-DGUAAG3T.js";var L=o(()=>{"use strict";e();g();r()});export{L as a};

package/dist/cli/chunks/client-3GECKJWB.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{c as a,d as b,e as c}from"./chunk-GIUAY4KY.js";import"./chunk-ME4LVBTT.js";import"./chunk-UIYK4Y7K.js";import"./chunk-DGUAAG3T.js";c();export{a as VibiumClientImpl,b as VibiumClientProvider};

package/dist/cli/chunks/cross-domain-router-UYONX76U.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b}from"./chunk-KFAN5FZM.js";import"./chunk-2L3BCZ2F.js";import"./chunk-PZF5ISGS.js";import"./chunk-G7L5Y47C.js";import"./chunk-DGUAAG3T.js";export{a as CrossDomainEventRouter,b as createCrossDomainRouter};

package/dist/cli/chunks/dream-Z5QDQJSM.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");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}from"./chunk-M2RO34V7.js";import"./chunk-HJJTLOWZ.js";import"./chunk-JOGKQTCU.js";import"./chunk-7732TLNA.js";import"./chunk-3PZ2QLLR.js";import"./chunk-NWZYULZE.js";import"./chunk-KJB6JJXT.js";import"./chunk-S2KGZJHS.js";import"./chunk-NWGIQCCO.js";import"./chunk-UAAJ3BSW.js";import"./chunk-ZCRLZINE.js";import"./chunk-TJOZPPZR.js";import"./chunk-4EQWK6P6.js";import"./chunk-Y3X5OS5F.js";import"./chunk-WSZVSDD5.js";import"./chunk-KVN4FTPT.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{t as BUILT_IN_STRATEGIES,b as ConceptGraph,g as DEFAULT_ACTIVATION_CONFIG,a as DEFAULT_CONCEPT_GRAPH_CONFIG,m as DEFAULT_DREAM_CONFIG,q as DEFAULT_DREAM_SCHEDULER_CONFIG,i as DEFAULT_INSIGHT_CONFIG,k as DEFAULT_VALIDATION_THRESHOLDS,n as DreamEngine,r as DreamScheduler,f as HISTORY_TRIM_TARGET_RATIO,j as InsightGenerator,d as MAX_ACTIVATION_HISTORY_ENTRIES,e as MAX_COACTIVATION_ENTRIES,l as RVCOWBranchManager,u as SpeculativeDreamer,h as SpreadingActivation,c as createConceptGraph,o as createDreamEngine,s as createDreamScheduler,p as default};

package/dist/cli/chunks/esm-node-FOI5Y3O2.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c,d,e,f,g,h,i}from"./chunk-G7L5Y47C.js";import"./chunk-DGUAAG3T.js";export{h as NIL,d as parse,b as stringify,c as v1,e as v3,f as v4,g as v5,a as validate,i as version};

package/dist/cli/chunks/eval-FM3BLVFO.js

@@ -1,15 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a as S}from"./chunk-DNFDSDKE.js";import{a as w,b as g}from"./chunk-53KLVMAT.js";import{a as y}from"./chunk-MFGYMNNY.js";import"./chunk-WUYJYRBY.js";import"./chunk-FNKKENLS.js";import"./chunk-2L3BCZ2F.js";import"./chunk-KJB6JJXT.js";import"./chunk-NWGIQCCO.js";import"./chunk-UAAJ3BSW.js";import"./chunk-WSZVSDD5.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-4RDTR2OF.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";import{Command as E}from"commander";import s from"chalk";import*as d from"fs";import*as p from"path";var f=["accessibility-testing","security-testing","performance-testing","chaos-engineering-resilience","contract-testing"],b=["risk-based-testing","test-design-techniques","regression-testing","test-data-management","localization-testing"],T=["quality-metrics","refactoring-patterns","iterative-loop","stream-chain"];function x(t){switch(t){case 3:return f;case 2:return[...f,...b];case 1:return[...f,...b,...T];default:return f}}var h="";function F(t){let e=Math.round(t.completedTasks/t.totalTasks*100),o=k(t.elapsedMs),l=k(t.estimatedRemainingMs),a=O(e,30),i=`${s.cyan("Progress:")} ${a} ${e}% | ${s.green(t.completedTasks)}/${t.totalTasks} tasks | ${s.yellow(t.failedTasks)} failed | ${s.blue(t.activeWorkers)} workers | ${o} elapsed, ~${l} remaining`;h&&process.stdout.write("\r"+" ".repeat(h.length)+"\r"),process.stdout.write(i),h=i}function O(t,e){let o=Math.round(t/100*e),l=e-o;return s.green("=".repeat(o))+s.gray("-".repeat(l))}function k(t){if(t<1e3)return`${t}ms`;if(t<6e4)return`${(t/1e3).toFixed(1)}s`;let e=Math.floor(t/6e4),o=Math.round(t%6e4/1e3);return`${e}m ${o}s`}function L(t){let e=[];e.push(""),e.push(s.bold(`Eval Results: ${t.skill}`)+s.dim(` (${t.model})`)),e.push("=".repeat(60));let o=t.passed?s.green("PASSED"):s.red("FAILED");e.push(`Status: ${o}`),e.push(""),e.push(s.bold("Metrics:")),e.push(`  Total Tests:    ${t.totalTests}`),e.push(`  Passed:         ${s.green(t.passedTests)}`),e.push(`  Failed:         ${s.red(t.failedTests)}`),e.push(`  Skipped:        ${s.yellow(t.skippedTests)}`),e.push(`  Pass Rate:      ${(t.passRate*100).toFixed(1)}%`),e.push(`  Reasoning Avg:  ${(t.avgReasoningQuality*100).toFixed(1)}%`),e.push(""),e.push(s.bold("Performance:")),e.push(`  Duration:       ${k(t.totalDurationMs)}`),e.push(`  Workers Used:   ${t.workersUsed}`),e.push(`  Parallel Speedup: ${s.cyan(t.parallelSpeedup.toFixed(2)+"x")}`),e.push("");let l=t.testResults.filter(a=>!a.passed);if(l.length>0){e.push(s.bold.red("Failed Tests:"));for(let a of l.slice(0,5))e.push(`  ${s.red("x")} ${a.testId}`),a.error&&e.push(`    ${s.dim(a.error)}`);l.length>5&&e.push(`  ... and ${l.length-5} more`),e.push("")}return e.join(`
-`)}function M(t){let e=[];e.push("# Skill Evaluation Report"),e.push(""),e.push(`Generated: ${new Date().toISOString()}`),e.push(""),e.push("## Summary"),e.push(""),e.push("| Skill | Model | Pass Rate | Duration | Speedup | Status |"),e.push("|-------|-------|-----------|----------|---------|--------|");for(let o of t){let l=o.passed?"PASSED":"FAILED";e.push(`| ${o.skill} | ${o.model} | ${(o.passRate*100).toFixed(1)}% | ${k(o.totalDurationMs)} | ${o.parallelSpeedup.toFixed(2)}x | ${l} |`)}e.push(""),e.push("## Detailed Results"),e.push("");for(let o of t){e.push(`### ${o.skill} (${o.model})`),e.push(""),e.push(`- **Total Tests:** ${o.totalTests}`),e.push(`- **Passed:** ${o.passedTests}`),e.push(`- **Failed:** ${o.failedTests}`),e.push(`- **Pass Rate:** ${(o.passRate*100).toFixed(1)}%`),e.push(`- **Avg Reasoning Quality:** ${(o.avgReasoningQuality*100).toFixed(1)}%`),e.push("");let l=o.testResults.filter(a=>!a.passed);if(l.length>0){e.push("**Failed Tests:**"),e.push("");for(let a of l)e.push(`- \`${a.testId}\`: ${a.error||"No error message"}`);e.push("")}}return e.join(`
-`)}async function $(){let t=await S({sqlite:{dbPath:".agentic-qe/eval-runner.db"},hnsw:{M:16,efConstruction:100,efSearch:50}});await t.initialize();let e=y(t);return{runner:g(e),learner:e,reasoningBank:t}}async function C(t){console.log(s.bold(`
-AQE Parallel Eval Runner`)),console.log(s.dim(`ADR-056 Phase 5: Worker Pool Pattern
-`));let{runner:e,reasoningBank:o}=await $();try{let l={...w,maxWorkers:t.parallel?t.workers:1,batchSize:t.batchSize,timeout:t.timeout,retryFailedTests:t.retry},a=g(y(o),l);t.verbose&&a.onProgress(F),console.log(s.cyan(`Running eval suite: ${t.skill}`)),console.log(s.cyan(`Model: ${t.model}`)),console.log(s.cyan(`Mode: ${t.parallel?`Parallel (${t.workers} workers)`:"Sequential"}`)),console.log("");let i=await a.runEvalParallel(t.skill,t.model);if(h&&console.log(""),console.log(L(i)),t.output){let r=p.resolve(t.output);d.writeFileSync(r,JSON.stringify(i,null,2)),console.log(s.green(`Results saved to: ${r}`))}process.exit(i.passed?0:1)}finally{await o.dispose()}}async function A(t){console.log(s.bold(`
-AQE Parallel Eval Runner - Multi-Skill`)),console.log(s.dim(`ADR-056 Phase 5: Worker Pool Pattern
-`));let{runner:e,learner:o,reasoningBank:l}=await $();try{let a=x(t.skillsTier),i=t.models.split(",").map(u=>u.trim());console.log(s.cyan(`Running evals for ${a.length} skills`)),console.log(s.cyan(`Models: ${i.join(", ")}`)),console.log(s.cyan(`Mode: ${t.parallel?`Parallel (${t.workers} workers)`:"Sequential"}`)),console.log("");let r={...w,maxWorkers:t.parallel?t.workers:1},m=await g(o,r).runMultipleEvalsParallel(a,i),c=[];for(let u of m.values())c.push(...u);console.log(s.bold(`
-Eval Summary`)),console.log("=".repeat(60));let R=0,v=0;for(let u of c){let P=u.passed?s.green("v"):s.red("x");console.log(`${P} ${u.skill} (${u.model}): ${(u.passRate*100).toFixed(1)}% (${u.parallelSpeedup.toFixed(2)}x speedup)`),u.passed?R++:v++}if(console.log(""),console.log(`Total: ${s.green(R)} passed, ${s.red(v)} failed`),t.output){let u=p.resolve(t.output);(u.endsWith(".md")?"markdown":"json")==="markdown"?d.writeFileSync(u,M(c)):d.writeFileSync(u,JSON.stringify(c,null,2)),console.log(s.green(`
-Report saved to: ${u}`))}process.exit(v===0?0:1)}finally{await l.dispose()}}async function D(t){let{learner:e,reasoningBank:o}=await $();try{let l=await e.getSkillConfidence(t);if(!l){console.log(s.yellow(`No validation history found for skill: ${t}`));return}if(console.log(s.bold(`
-Skill Validation Status: ${t}`)),console.log("=".repeat(50)),console.log(`Average Score: ${(l.avgScore*100).toFixed(1)}%`),console.log(`Trend: ${l.trend}`),console.log(`Validation Count: ${l.outcomes.length}`),console.log(`Last Updated: ${l.lastUpdated}`),l.byLevel){console.log(`
-By Validation Level:`);for(let[i,r]of Object.entries(l.byLevel))r!==void 0&&console.log(`  ${i}: ${(r*100).toFixed(1)}%`)}let a=await e.getCrossModelAnalysis(t);if(a&&(console.log(`
-Cross-Model Analysis:`),console.log(`  Variance: ${(a.variance*100).toFixed(2)}%`),console.log(`  Has Anomalies: ${a.hasAnomalies}`),a.models)){console.log(`
-  Model Performance:`);for(let[i,r]of Object.entries(a.models))console.log(`    ${i}: ${(r.passRate*100).toFixed(1)}% pass rate (${r.sampleCount} samples)`)}}finally{await o.dispose()}}async function I(t){let{learner:e,reasoningBank:o}=await $();try{let l=await e.queryValidationPatterns(t.skill,50);if(l.length===0){console.log(s.yellow(`No validation patterns found for skill: ${t.skill}`));return}let a=await e.extractLearnedPatterns(t.skill),i=await e.getValidationTrends(t.skill),r={skill:t.skill,generated:new Date().toISOString(),patternCount:l.length,learnedPatterns:a,trends:i};if(t.format==="json"){let n=JSON.stringify(r,null,2);t.output?(d.writeFileSync(p.resolve(t.output),n),console.log(s.green(`Report saved to: ${t.output}`))):console.log(n)}else{let n=[];if(n.push(`# Validation Report: ${t.skill}`),n.push(""),n.push(`Generated: ${r.generated}`),n.push(""),n.push("## Overview"),n.push(`- Patterns Collected: ${r.patternCount}`),n.push(`- Learned Patterns: ${r.learnedPatterns.length}`),n.push(""),r.trends&&(n.push("## Trends"),n.push(`- Overall: ${r.trends.overall}`),n.push(`- Recent Pass Rate: ${(r.trends.recentPassRate*100).toFixed(1)}%`),n.push("")),r.learnedPatterns.length>0){n.push("## Learned Patterns"),n.push("");for(let c of r.learnedPatterns)n.push(`### ${c.category}`),n.push(`- Confidence: ${(c.confidence*100).toFixed(1)}%`),n.push(`- Observations: ${c.observationCount}`),n.push(`- Models: ${c.models.join(", ")}`),n.push("")}let m=n.join(`
-`);t.output?(d.writeFileSync(p.resolve(t.output),m),console.log(s.green(`Report saved to: ${t.output}`))):console.log(m)}}finally{await o.dispose()}}function G(){let t=new E("eval").description("Run skill evaluation suites in parallel");return t.command("run").description("Run eval suite for a single skill").requiredOption("-s, --skill <skill>","Skill name to evaluate").requiredOption("-m, --model <model>","Model to use (e.g., claude-3.5-sonnet)").option("-p, --parallel","Enable parallel execution",!1).option("-w, --workers <n>","Number of parallel workers",parseInt,5).option("-b, --batch-size <n>","Test cases per batch",parseInt,4).option("-t, --timeout <ms>","Timeout per test case (ms)",parseInt,3e4).option("--no-retry","Disable retry of failed tests").option("-o, --output <path>","Output file path for results").option("-v, --verbose","Show progress during execution",!1).action(async e=>{await C({skill:e.skill,model:e.model,parallel:e.parallel,workers:e.workers,batchSize:e.batchSize,timeout:e.timeout,retry:e.retry!==!1,output:e.output,verbose:e.verbose})}),t.command("run-all").description("Run eval suites for multiple skills").option("--skills-tier <tier>","Skill tier (1=all, 2=P0+P1, 3=P0 only)",parseInt,3).option("--models <models>","Comma-separated models to test","claude-3.5-sonnet").option("-p, --parallel","Enable parallel execution",!0).option("-w, --workers <n>","Number of parallel workers",parseInt,5).option("-o, --output <path>","Output file path for report").option("-v, --verbose","Show progress during execution",!1).action(async e=>{await A({skillsTier:e.skillsTier,models:e.models,parallel:e.parallel,workers:e.workers,output:e.output,verbose:e.verbose})}),t.command("status").description("Show validation status for a skill").requiredOption("-s, --skill <skill>","Skill name").action(async e=>{await D(e.skill)}),t.command("report").description("Generate validation report for a skill").requiredOption("-s, --skill <skill>","Skill name").option("-f, --format <format>","Output format (json|markdown)","markdown").option("-o, --output <path>","Output file path").action(async e=>{await I({skill:e.skill,format:e.format,output:e.output})}),t}export{G as createEvalCommand};

package/dist/cli/chunks/hnsw-adapter-NOAQ76QF.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,c as b}from"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";b();export{a as HnswAdapter};

package/dist/cli/chunks/hnsw-index-KDIZQRSB.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c,d}from"./chunk-RQFC7Q33.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-JNLKLC7G.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{a as DEFAULT_HNSW_CONFIG,b as HNSWIndex,d as benchmarkHNSW,c as createHNSWIndex};

package/dist/cli/chunks/impact-analyzer-VDXJBWFG.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a}from"./chunk-GM5LIYLQ.js";import"./chunk-GOTVYSX5.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-DGUAAG3T.js";export{a as ImpactAnalyzerService};

package/dist/cli/chunks/init-wizard-Q6DEACKX.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{a,b,c,d}from"./chunk-CGC63MGF.js";import"./chunk-HEQ23HBE.js";import"./chunk-BFCMOMOF.js";import"./chunk-XH774Q6Z.js";import"./chunk-L4IWZXSC.js";import"./chunk-DMZ2XOEW.js";import"./chunk-UIYK4Y7K.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-DGUAAG3T.js";export{a as InitOrchestrator,b as createInitOrchestrator,d as formatInitResult,c as quickInit};

package/dist/cli/chunks/kernel-YNQOGTM5.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{v as a,w as b}from"./chunk-6EZY2QVM.js";import"./chunk-JWR3NNFA.js";import"./chunk-J5YOXLBD.js";import"./chunk-I3P5RZX7.js";import"./chunk-UG2J5MTY.js";import"./chunk-G2KQUEGY.js";import"./chunk-KHUFPJE4.js";import"./chunk-5DWTU2PN.js";import"./chunk-36YZ2UF7.js";import"./chunk-VDGWDDBC.js";import"./chunk-D6APBW66.js";import"./chunk-FNKKENLS.js";import"./chunk-2L3BCZ2F.js";import"./chunk-GIUAY4KY.js";import"./chunk-M2RO34V7.js";import"./chunk-HJJTLOWZ.js";import"./chunk-JOGKQTCU.js";import"./chunk-7732TLNA.js";import"./chunk-3PZ2QLLR.js";import"./chunk-NWZYULZE.js";import"./chunk-KJB6JJXT.js";import"./chunk-S2KGZJHS.js";import"./chunk-NWGIQCCO.js";import"./chunk-UAAJ3BSW.js";import"./chunk-ZCRLZINE.js";import"./chunk-SUXUCW5I.js";import"./chunk-TJOZPPZR.js";import"./chunk-VQB5SLXZ.js";import"./chunk-6BK2S2WE.js";import"./chunk-ME4LVBTT.js";import"./chunk-5MWNXDQW.js";import"./chunk-RQFC7Q33.js";import"./chunk-4EQWK6P6.js";import"./chunk-GM5LIYLQ.js";import"./chunk-GOTVYSX5.js";import"./chunk-ZU5VHW6P.js";import"./chunk-SY74WSBN.js";import"./chunk-Y3X5OS5F.js";import"./chunk-WSZVSDD5.js";import"./chunk-KVN4FTPT.js";import"./chunk-ZFZX2NA3.js";import"./chunk-UUVTLYUK.js";import"./chunk-4RDTR2OF.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-VU4QET3C.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-G7L5Y47C.js";import"./chunk-A2TK52B2.js";import"./chunk-LWHM5BWD.js";import"./chunk-J3XLOHWB.js";import"./chunk-JN7MSUGK.js";import"./chunk-JNLKLC7G.js";import"./chunk-WE7KIXR6.js";import"./chunk-XNXOW7BQ.js";import"./chunk-JLBUS4PR.js";import"./chunk-DJLRRCR6.js";import"./chunk-5HUAD4UU.js";import"./chunk-DGUAAG3T.js";export{a as QEKernelImpl,b as createKernel};

package/dist/cli/chunks/knowledge-graph-GOZG3YOM.js

@@ -1,2 +0,0 @@
-import{createRequire as __cr}from"module";const require=__cr(import.meta.url);if(process.argv.includes('--version')||process.argv.includes('-v')){console.log("3.9.14");process.exit(0)}
-import{g as a}from"./chunk-GOTVYSX5.js";import"./chunk-STYNOKOM.js";import"./chunk-GUEK5DOK.js";import"./chunk-DMZ2XOEW.js";import"./chunk-PZF5ISGS.js";import"./chunk-UIYK4Y7K.js";import"./chunk-DGUAAG3T.js";export{a as KnowledgeGraphService};