@sienklogic/plan-build-run 2.21.0 → 2.21.2

package/plugins/pbr/scripts/audit-dimensions.js

@@ -0,0 +1,556 @@
+'use strict';
+
+/**
+ * Audit Dimension Registry
+ *
+ * Central registry of all audit dimensions used by /pbr:audit.
+ * Each dimension has a hybrid code+slug ID (e.g., SI-01:skill-template-refs),
+ * category, severity, description, phase origin, and source field.
+ *
+ * Dimension schema:
+ *   id            - Hybrid code+slug (e.g., "SI-01:skill-template-refs")
+ *   code          - Short code for CLI/reports (e.g., "SI-01")
+ *   slug          - Slug for CLI (e.g., "skill-template-refs")
+ *   category      - Config category key (e.g., "self_integrity")
+ *   categoryCode  - Category prefix (e.g., "SI")
+ *   severity      - Default severity: "error" | "warning" | "info"
+ *   description   - Human-readable one-liner
+ *   phaseOrigin   - ROADMAP phase number that implements this dimension
+ *   source        - "builtin" (future: "custom" for user-defined dimensions)
+ *   thresholdKey  - Config audit.thresholds key if applicable, else null
+ */
+
+// ---------------------------------------------------------------------------
+// Category codes -> config.json audit.categories keys
+// ---------------------------------------------------------------------------
+const CATEGORIES = {
+  AC: 'audit_config',
+  SI: 'self_integrity',
+  IH: 'infrastructure',
+  EF: 'error_analysis',
+  WC: 'workflow_compliance',
+  BC: 'behavioral_compliance',
+  SQ: 'session_quality',
+  FV: 'feature_verification',
+  QM: 'quality_metrics',
+};
+
+// ---------------------------------------------------------------------------
+// Presets -> arrays of category keys
+// ---------------------------------------------------------------------------
+const PRESETS = {
+  minimal: ['self_integrity', 'infrastructure'],
+  standard: [
+    'self_integrity',
+    'infrastructure',
+    'error_analysis',
+    'workflow_compliance',
+  ],
+  comprehensive: [
+    'audit_config',
+    'self_integrity',
+    'infrastructure',
+    'error_analysis',
+    'workflow_compliance',
+    'behavioral_compliance',
+    'session_quality',
+    'feature_verification',
+    'quality_metrics',
+  ],
+  custom: [],
+};
+
+// ---------------------------------------------------------------------------
+// Helper to build a dimension entry
+// ---------------------------------------------------------------------------
+function dim(code, slug, category, categoryCode, severity, description, phaseOrigin, thresholdKey) {
+  return {
+    id: `${code}:${slug}`,
+    code,
+    slug,
+    category,
+    categoryCode,
+    severity,
+    description,
+    phaseOrigin,
+    source: 'builtin',
+    thresholdKey: thresholdKey || null,
+  };
+}
+
+// ---------------------------------------------------------------------------
+// DIMENSIONS array — all ~88 dimensions from ROADMAP phases 1-9
+// ---------------------------------------------------------------------------
+const DIMENSIONS = [
+
+  // =========================================================================
+  // AC — Audit Config (Phase 1) — 1 dimension
+  // =========================================================================
+  dim('AC-01', 'config-schema-completeness', 'audit_config', 'AC', 'warning',
+    'Config schema completeness — detect config vs schema drift', 1, null),
+
+  // =========================================================================
+  // SI — Self-Integrity (Phase 2) — 15 dimensions
+  // =========================================================================
+  dim('SI-01', 'skill-template-refs', 'self_integrity', 'SI', 'error',
+    'Skill template reference validation — all ${CLAUDE_SKILL_DIR}/templates/ refs resolve', 2, null),
+  dim('SI-02', 'skill-shared-fragment-refs', 'self_integrity', 'SI', 'error',
+    'Skill shared fragment reference validation — all skills/shared/*.md refs exist', 2, null),
+  dim('SI-03', 'skill-reference-file-links', 'self_integrity', 'SI', 'error',
+    'Skill reference file link validation — all references/*.md links exist', 2, null),
+  dim('SI-04', 'skill-agent-type-refs', 'self_integrity', 'SI', 'error',
+    'Skill to agent type reference validation — all subagent_type: "pbr:X" map to existing agents', 2, null),
+  dim('SI-05', 'skill-agent-completion-markers', 'self_integrity', 'SI', 'error',
+    'Skill/agent completion marker contract — skills check same markers agents output', 2, null),
+  dim('SI-06', 'agent-frontmatter-validity', 'self_integrity', 'SI', 'error',
+    'Agent frontmatter validity — name, description, model, tools present', 2, null),
+  dim('SI-07', 'agent-tool-list-accuracy', 'self_integrity', 'SI', 'warning',
+    'Agent tool list accuracy — listed tools match actual needs', 2, null),
+  dim('SI-08', 'hook-script-existence', 'self_integrity', 'SI', 'error',
+    'Hook script existence — all hooks.json entries point to existing scripts', 2, null),
+  dim('SI-09', 'pretooluse-stdout-compliance', 'self_integrity', 'SI', 'error',
+    'PreToolUse stdout compliance — all PreToolUse hooks emit JSON on stdout', 2, null),
+  dim('SI-10', 'command-skill-mapping', 'self_integrity', 'SI', 'error',
+    'Command to skill mapping validity — all commands point to existing skills', 2, null),
+  dim('SI-11', 'config-schema-code-consistency', 'self_integrity', 'SI', 'warning',
+    'Config schema vs code consistency — settings in code exist in schema', 2, null),
+  dim('SI-12', 'plugin-manifest-version-sync', 'self_integrity', 'SI', 'error',
+    'Plugin manifest version sync — plugin.json matches package.json', 2, null),
+  dim('SI-13', 'dispatch-chain-completeness', 'self_integrity', 'SI', 'warning',
+    'Dispatch chain completeness — dispatchers call all documented sub-hooks', 2, null),
+  dim('SI-14', 'critical-marker-coverage', 'self_integrity', 'SI', 'warning',
+    'CRITICAL marker coverage — known failure-prone steps have CRITICAL/STOP markers', 2, null),
+  dim('SI-15', 'cross-platform-path-safety', 'self_integrity', 'SI', 'warning',
+    'Cross-platform path safety — no hardcoded / or \\, use path.join', 2, null),
+
+  // =========================================================================
+  // IH — Infrastructure Health (Phase 3) — 10 dimensions
+  // =========================================================================
+  dim('IH-01', 'hook-server-health', 'infrastructure', 'IH', 'error',
+    'Hook server health — running/crashed/process-fallback detection', 3, null),
+  dim('IH-02', 'dashboard-health', 'infrastructure', 'IH', 'warning',
+    'Dashboard health — running if auto_launch enabled', 3, null),
+  dim('IH-03', 'hook-exec-performance', 'infrastructure', 'IH', 'warning',
+    'Hook execution performance profiling — flag >500ms hooks', 3, 'hook_performance_ms'),
+  dim('IH-04', 'stale-file-detection', 'infrastructure', 'IH', 'warning',
+    'Stale file detection — .active-skill, .auto-next, .context-tracker from crashed sessions', 3, null),
+  dim('IH-05', 'plugin-cache-freshness', 'infrastructure', 'IH', 'info',
+    'Plugin cache freshness — cached version matches local/published version', 3, null),
+  dim('IH-06', 'config-schema-validation', 'infrastructure', 'IH', 'error',
+    'Config schema validation — config.json valid against schema, no unknown keys', 3, null),
+  dim('IH-07', 'log-rotation-health', 'infrastructure', 'IH', 'info',
+    'Log rotation health — cleanOldHookLogs working, old files removed after 30 days', 3, null),
+  dim('IH-08', 'disk-usage-tracking', 'infrastructure', 'IH', 'info',
+    'Disk usage tracking — .planning/ size, milestone archive growth', 3, null),
+  dim('IH-09', 'dispatch-chain-coverage', 'infrastructure', 'IH', 'warning',
+    'Dispatch chain coverage — all 13 sub-hooks actually executing', 3, null),
+  dim('IH-10', 'log-source-separation', 'infrastructure', 'IH', 'info',
+    'Log source separation quality — test vs session data in events-*.jsonl', 3, null),
+
+  // =========================================================================
+  // EF — Error & Failure Analysis (Phase 4) — 7 dimensions
+  // =========================================================================
+  dim('EF-01', 'tool-failure-rate', 'error_analysis', 'EF', 'error',
+    'Tool failure rate analysis — PostToolUseFailure events by tool type', 4, null),
+  dim('EF-02', 'agent-failure-timeout', 'error_analysis', 'EF', 'error',
+    'Agent failure/timeout detection — missing completion markers, null durations', 4, null),
+  dim('EF-03', 'hook-false-positive', 'error_analysis', 'EF', 'warning',
+    'Hook false positive analysis — blocks on legitimate writes', 4, null),
+  dim('EF-04', 'hook-false-negative', 'error_analysis', 'EF', 'warning',
+    'Hook false negative detection — bad actions not blocked', 4, null),
+  dim('EF-05', 'retry-repetition-pattern', 'error_analysis', 'EF', 'warning',
+    'Retry/repetition pattern detection — same tool called 3+ times consecutively', 4, null),
+  dim('EF-06', 'cross-session-interference', 'error_analysis', 'EF', 'error',
+    'Cross-session interference detection — multiple sessions, .active-skill conflicts', 4, null),
+  dim('EF-07', 'session-cleanup-verification', 'error_analysis', 'EF', 'warning',
+    'Session cleanup verification — session-cleanup.js fires and cleans up properly', 4, null),
+  dim('EF-08', 'incident-pattern-analysis', 'error_analysis', 'EF', 'warning',
+    'Analyzes incident journal for recurring patterns, high-severity clusters, and auto-fix effectiveness', 46, null),
+
+  // =========================================================================
+  // WC — Workflow Compliance (Phase 5) — 12 dimensions
+  // =========================================================================
+  dim('WC-01', 'ci-verify-after-push', 'workflow_compliance', 'WC', 'warning',
+    'CI verification after push — git push followed by gh run list check', 5, null),
+  dim('WC-02', 'state-file-integrity', 'workflow_compliance', 'WC', 'error',
+    'State file integrity — STATE.md matches disk, phase statuses accurate', 5, null),
+  dim('WC-03', 'state-frontmatter-integrity', 'workflow_compliance', 'WC', 'error',
+    'STATE.md frontmatter integrity — valid YAML, required fields present', 5, null),
+  dim('WC-04', 'roadmap-sync-validation', 'workflow_compliance', 'WC', 'error',
+    'ROADMAP sync validation — ROADMAP.md matches actual phase directories', 5, null),
+  dim('WC-05', 'planning-artifact-completeness', 'workflow_compliance', 'WC', 'warning',
+    'Planning artifact completeness — built phases have SUMMARY.md + VERIFICATION.md', 5, null),
+  dim('WC-06', 'artifact-format-validation', 'workflow_compliance', 'WC', 'warning',
+    'Artifact format validation — SUMMARY.md required fields, PLAN.md has task blocks', 5, null),
+  dim('WC-07', 'compaction-quality', 'workflow_compliance', 'WC', 'info',
+    'Compaction quality tracking — STATE.md preserved through compaction events', 5, null),
+  dim('WC-08', 'naming-convention-compliance', 'workflow_compliance', 'WC', 'warning',
+    'Naming convention compliance — PLAN-{NN}.md format, not random names', 5, null),
+  dim('WC-09', 'commit-pattern-validation', 'workflow_compliance', 'WC', 'warning',
+    'Commit command pattern validation — no heredoc, proper -m format', 5, null),
+  dim('WC-10', 'model-selection-compliance', 'workflow_compliance', 'WC', 'info',
+    'Model selection compliance — agents using configured models', 5, null),
+  dim('WC-11', 'git-branching-compliance', 'workflow_compliance', 'WC', 'info',
+    'Git branching compliance — phase branches created per git.branching setting', 5, null),
+  dim('WC-12', 'test-health-baseline', 'workflow_compliance', 'WC', 'warning',
+    'Test health baseline tracking — new failures vs known failures', 5, null),
+
+  // =========================================================================
+  // BC — Behavioral Compliance (Phase 6) — 15 dimensions
+  // =========================================================================
+  dim('BC-01', 'skill-sequence-compliance', 'behavioral_compliance', 'BC', 'error',
+    'Skill sequence compliance — plan before build, build before verify', 6, null),
+  dim('BC-02', 'state-machine-transitions', 'behavioral_compliance', 'BC', 'error',
+    'State machine transition validation — planned>building>built>verified, no skips', 6, null),
+  dim('BC-03', 'pre-condition-verification', 'behavioral_compliance', 'BC', 'error',
+    'Pre-condition verification — PLAN.md exists before build, SUMMARY.md before verify', 6, null),
+  dim('BC-04', 'post-condition-verification', 'behavioral_compliance', 'BC', 'warning',
+    'Post-condition verification — SUMMARY.md after build, VERIFICATION.md after verify', 6, null),
+  dim('BC-05', 'orchestrator-budget-discipline', 'behavioral_compliance', 'BC', 'warning',
+    'Orchestrator budget discipline — stayed under budget, no executor file reads', 6, null),
+  dim('BC-06', 'artifact-creation-order', 'behavioral_compliance', 'BC', 'warning',
+    'Artifact creation order — PLAN>commits>SUMMARY>VERIFICATION in order', 6, null),
+  dim('BC-07', 'critical-marker-compliance', 'behavioral_compliance', 'BC', 'warning',
+    'CRITICAL marker compliance — skills followed their CRITICAL/STOP markers', 6, null),
+  dim('BC-08', 'gate-compliance', 'behavioral_compliance', 'BC', 'warning',
+    'Gate compliance — configured gates respected in autonomous mode', 6, null),
+  dim('BC-09', 'enforce-workflow-advisory', 'behavioral_compliance', 'BC', 'info',
+    'Enforce-PBR-workflow advisory tracking — enforcement hook flags and responses', 6, null),
+  dim('BC-10', 'unmanaged-commit-detection', 'behavioral_compliance', 'BC', 'warning',
+    'Unmanaged commit detection — commits made outside PBR skill context', 6, null),
+  dim('BC-11', 'context-delegation-threshold', 'behavioral_compliance', 'BC', 'warning',
+    'Context delegation threshold — subagents spawned when context exceeded cap', 6, null),
+  dim('BC-12', 'skill-self-read-prevention', 'behavioral_compliance', 'BC', 'info',
+    'Skill self-read prevention — no skill wasted tokens reading its own SKILL.md', 6, null),
+  dim('BC-13', 'hook-output-effectiveness', 'behavioral_compliance', 'BC', 'info',
+    'Hook output effectiveness — LLM actually followed hook advisories/warnings', 6, null),
+  dim('BC-14', 'agent-scope-compliance', 'behavioral_compliance', 'BC', 'error',
+    'Agent scope compliance — only modified files listed in plan files_modified', 6, null),
+  dim('BC-15', 'agent-plan-adherence', 'behavioral_compliance', 'BC', 'warning',
+    'Agent plan adherence — tasks_completed matches plan, executor self-check results', 6, null),
+
+  // =========================================================================
+  // SQ — Session Quality & UX (Phase 7) — 10 dimensions
+  // =========================================================================
+  dim('SQ-01', 'session-start-quality', 'session_quality', 'SQ', 'info',
+    'Session start quality — progress-tracker briefing injected and useful', 7, null),
+  dim('SQ-02', 'briefing-freshness', 'session_quality', 'SQ', 'info',
+    'Briefing freshness/size — stale or bloated session start injection', 7, null),
+  dim('SQ-03', 'session-duration-cost', 'session_quality', 'SQ', 'info',
+    'Session duration/cost analysis — flag bloated sessions', 7, null),
+  dim('SQ-04', 'skill-routing-accuracy', 'session_quality', 'SQ', 'warning',
+    'Skill routing accuracy — /pbr:do routing to correct skill', 7, null),
+  dim('SQ-05', 'memory-update-tracking', 'session_quality', 'SQ', 'info',
+    'Memory update tracking — auto-memory saves when learning new info', 7, null),
+  dim('SQ-06', 'convention-detection-monitoring', 'session_quality', 'SQ', 'info',
+    'Convention detection monitoring — convention_memory capturing patterns', 7, null),
+  dim('SQ-07', 'user-frustration-signals', 'session_quality', 'SQ', 'warning',
+    'User frustration signals — repeated commands, corrections, negative language', 7, null),
+  dim('SQ-08', 'satisfaction-signals', 'session_quality', 'SQ', 'info',
+    'Satisfaction signals — positive vs negative language ratio', 7, null),
+  dim('SQ-09', 'skill-escalation-patterns', 'session_quality', 'SQ', 'info',
+    'Skill escalation patterns — quick>debug, quick>plan, task harder than expected', 7, null),
+  dim('SQ-10', 'notification-quality', 'session_quality', 'SQ', 'info',
+    'Notification quality — log-notification.js notifications useful or noise', 7, null),
+
+  // =========================================================================
+  // FV — Feature Verification (Phase 8) — 13 dimensions
+  // =========================================================================
+  dim('FV-01', 'architecture-guard-activity', 'feature_verification', 'FV', 'warning',
+    'Architecture guard activity — dependency violation warnings generated', 8, null),
+  dim('FV-02', 'dependency-break-detection', 'feature_verification', 'FV', 'warning',
+    'Dependency break detection activity — stale plans flagged on upstream changes', 8, null),
+  dim('FV-03', 'security-scanning-activity', 'feature_verification', 'FV', 'info',
+    'Security scanning activity — OWASP scans run during builds', 8, null),
+  dim('FV-04', 'trust-tracking-activity', 'feature_verification', 'FV', 'info',
+    'Trust tracking activity — trust scores updated, influencing verification depth', 8, null),
+  dim('FV-05', 'learnings-system-activity', 'feature_verification', 'FV', 'info',
+    'Learnings system activity — LEARNINGS.md written and consumed by planners', 8, null),
+  dim('FV-06', 'intel-system-activity', 'feature_verification', 'FV', 'info',
+    'Intel system activity — arch.md updated, injected at session start', 8, null),
+  dim('FV-07', 'auto-continue-chain', 'feature_verification', 'FV', 'warning',
+    'Auto-continue chain verification — stop hook chaining correctly, no false triggers', 8, null),
+  dim('FV-08', 'negative-knowledge-tracking', 'feature_verification', 'FV', 'info',
+    'Negative knowledge tracking — failures recorded in .planning/negative-knowledge/', 8, null),
+  dim('FV-09', 'decision-journal-tracking', 'feature_verification', 'FV', 'info',
+    'Decision journal tracking — decisions journaled in .planning/decisions/', 8, null),
+  dim('FV-10', 'phase-boundary-enforcement', 'feature_verification', 'FV', 'warning',
+    'Phase boundary enforcement — cross-phase violations detected/missed', 8, null),
+  dim('FV-11', 'destructive-op-confirmation', 'feature_verification', 'FV', 'error',
+    'Destructive operation confirmation — destructive git ops confirmed per safety config', 8, null),
+  dim('FV-12', 'context-budget-accuracy', 'feature_verification', 'FV', 'info',
+    'Context-budget.json accuracy — bridged value matches reality', 8, null),
+  dim('FV-13', 'config-feature-coverage', 'feature_verification', 'FV', 'warning',
+    'Config feature coverage — % of enabled features with evidence of running', 8, null),
+
+  // =========================================================================
+  // QM — Quality Metrics & Trends (Phase 9) — 6 dimensions
+  // =========================================================================
+  dim('QM-01', 'session-degradation', 'quality_metrics', 'QM', 'warning',
+    'Session degradation tracking — error rate first half vs second half', 9, null),
+  dim('QM-02', 'throughput-metrics', 'quality_metrics', 'QM', 'info',
+    'Throughput metrics — tool calls, commits, agents spawned per session', 9, null),
+  dim('QM-03', 'baseline-comparison', 'quality_metrics', 'QM', 'warning',
+    'Baseline comparison — current audit vs previous audits for regression detection', 9, null),
+  dim('QM-04', 'error-correlation', 'quality_metrics', 'QM', 'info',
+    'Error correlation analysis — errors in one dimension predict errors in another', 9, null),
+  dim('QM-05', 'audit-self-validation', 'quality_metrics', 'QM', 'warning',
+    'Audit self-validation — audit agent checked all enabled dimensions', 9, null),
+  dim('QM-06', 'insights-coverage', 'quality_metrics', 'QM', 'info',
+    'Insights report freshness — /insights HTML report exists and is recent (within 30 days)', 49, null),
+];
+
+// ---------------------------------------------------------------------------
+// Lookup functions
+// ---------------------------------------------------------------------------
+
+/**
+ * Find a dimension by its full ID (e.g., "SI-01:skill-template-refs")
+ * or by its short code (e.g., "SI-01").
+ * @param {string} idOrCode
+ * @returns {object|undefined}
+ */
+function getDimensionById(idOrCode) {
+  return DIMENSIONS.find(
+    (d) => d.id === idOrCode || d.code === idOrCode
+  );
+}
+
+/**
+ * Find a dimension by its slug (e.g., "skill-template-refs").
+ * @param {string} slug
+ * @returns {object|undefined}
+ */
+function getDimensionBySlug(slug) {
+  return DIMENSIONS.find((d) => d.slug === slug);
+}
+
+/**
+ * Get all dimensions in a given category.
+ * @param {string} category - Config category key (e.g., "self_integrity")
+ * @returns {object[]}
+ */
+function getDimensionsByCategory(category) {
+  return DIMENSIONS.filter((d) => d.category === category);
+}
+
+/**
+ * Get the list of category keys for a given preset name.
+ * @param {string} presetName - "minimal" | "standard" | "comprehensive" | "custom"
+ * @returns {string[]}
+ */
+function getActivePresetCategories(presetName) {
+  return PRESETS[presetName] || [];
+}
+
+// ---------------------------------------------------------------------------
+// Resolve a dimension identifier (code like "SI-01" or slug like "skill-template-refs")
+// ---------------------------------------------------------------------------
+function resolveDimension(identifier) {
+  return getDimensionById(identifier) || getDimensionBySlug(identifier);
+}
+
+// ---------------------------------------------------------------------------
+// resolveDimensions(config, cliFlags) — compute the active dimension set
+//
+// Resolution order (LOCKED — additive composition):
+//   1. Start with preset (cliFlags.preset || config.audit.preset || "standard")
+//   2. Get base categories from PRESETS[preset]
+//   3. Apply config.audit.categories toggles
+//   4. Apply config.audit.overrides per-dimension toggles
+//   5. Apply cliFlags.dimension — ADD these dimensions
+//   6. Apply cliFlags.skip — REMOVE these dimensions
+//   7. If cliFlags.only is non-empty — OVERRIDE everything with exactly those
+//   8. Return sorted by code
+// ---------------------------------------------------------------------------
+
+/**
+ * Compute the final active dimension set.
+ * @param {object} config - Full config.json object (must have config.audit)
+ * @param {object} cliFlags - { preset?, dimension?, skip?, only? }
+ * @returns {object[]} Array of dimension objects sorted by code
+ */
+function resolveDimensions(config, cliFlags) {
+  const flags = cliFlags || {};
+  const audit = (config && config.audit) || {};
+
+  // Step 7 early: --only overrides everything
+  if (flags.only && flags.only.length > 0) {
+    const onlyMap = new Map();
+    for (const id of flags.only) {
+      const d = resolveDimension(id);
+      if (d) onlyMap.set(d.code, d);
+    }
+    return Array.from(onlyMap.values()).sort((a, b) => a.code.localeCompare(b.code));
+  }
+
+  // Step 1: Determine preset
+  const presetName = flags.preset || audit.preset || 'standard';
+  const presetCategories = PRESETS[presetName] || [];
+
+  // Step 2: Build active categories set from preset
+  const activeCategories = new Set(presetCategories);
+
+  // Step 3: Apply config.audit.categories toggles
+  const categoryToggles = audit.categories || {};
+  for (const [cat, enabled] of Object.entries(categoryToggles)) {
+    if (enabled === false) {
+      activeCategories.delete(cat);
+    } else if (enabled === true) {
+      activeCategories.add(cat);
+    }
+  }
+
+  // Collect dimensions from active categories
+  const activeMap = new Map();
+  for (const d of DIMENSIONS) {
+    if (activeCategories.has(d.category)) {
+      activeMap.set(d.code, d);
+    }
+  }
+
+  // Step 4: Apply config.audit.overrides per-dimension toggles
+  const overrides = audit.overrides || {};
+  for (const [id, enabled] of Object.entries(overrides)) {
+    if (enabled === false) {
+      // Try to resolve the identifier
+      const d = resolveDimension(id);
+      if (d) activeMap.delete(d.code);
+    } else if (enabled === true) {
+      const d = resolveDimension(id);
+      if (d) activeMap.set(d.code, d);
+    }
+  }
+
+  // Step 5: Apply cliFlags.dimension — ADD
+  if (flags.dimension && flags.dimension.length > 0) {
+    for (const id of flags.dimension) {
+      const d = resolveDimension(id);
+      if (d) activeMap.set(d.code, d);
+    }
+  }
+
+  // Step 6: Apply cliFlags.skip — REMOVE
+  if (flags.skip && flags.skip.length > 0) {
+    for (const id of flags.skip) {
+      const d = resolveDimension(id);
+      if (d) activeMap.delete(d.code);
+    }
+  }
+
+  return Array.from(activeMap.values()).sort((a, b) => a.code.localeCompare(b.code));
+}
+
+// ---------------------------------------------------------------------------
+// explainResolution(config, cliFlags) — human-readable trace of resolution
+// ---------------------------------------------------------------------------
+
+/**
+ * Return an array of strings explaining each resolution step.
+ * @param {object} config - Full config.json object
+ * @param {object} cliFlags - { preset?, dimension?, skip?, only? }
+ * @returns {string[]}
+ */
+function explainResolution(config, cliFlags) {
+  const flags = cliFlags || {};
+  const audit = (config && config.audit) || {};
+  const steps = [];
+
+  // --only overrides everything
+  if (flags.only && flags.only.length > 0) {
+    const resolved = flags.only.map(id => resolveDimension(id)).filter(Boolean);
+    steps.push(`--only override -> ${resolved.length} dimensions (ignoring preset/categories)`);
+    steps.push(`Final: ${resolved.length} active dimensions`);
+    return steps;
+  }
+
+  // Step 1: Preset
+  const presetName = flags.preset || audit.preset || 'standard';
+  const presetCategories = PRESETS[presetName] || [];
+  let baseDims = 0;
+  for (const cat of presetCategories) {
+    baseDims += getDimensionsByCategory(cat).length;
+  }
+  steps.push(`Preset '${presetName}' -> ${presetCategories.length} categories (${baseDims} dimensions)`);
+
+  // Step 2: Category toggles
+  const activeCategories = new Set(presetCategories);
+  const categoryToggles = audit.categories || {};
+  for (const [cat, enabled] of Object.entries(categoryToggles)) {
+    if (enabled === false && activeCategories.has(cat)) {
+      const removed = getDimensionsByCategory(cat).length;
+      activeCategories.delete(cat);
+      steps.push(`Category '${cat}' disabled by config -> -${removed} dimensions`);
+    } else if (enabled === true && !activeCategories.has(cat)) {
+      const added = getDimensionsByCategory(cat).length;
+      activeCategories.add(cat);
+      steps.push(`Category '${cat}' enabled by config -> +${added} dimensions`);
+    }
+  }
+
+  // Current count
+  let currentCount = 0;
+  const activeMap = new Map();
+  for (const d of DIMENSIONS) {
+    if (activeCategories.has(d.category)) {
+      activeMap.set(d.code, d);
+      currentCount++;
+    }
+  }
+
+  // Step 3: Overrides
+  const overrides = audit.overrides || {};
+  for (const [id, enabled] of Object.entries(overrides)) {
+    const d = resolveDimension(id);
+    if (!d) continue;
+    if (enabled === false && activeMap.has(d.code)) {
+      activeMap.delete(d.code);
+      steps.push(`Override ${d.code} disabled by config -> -1 dimension`);
+    } else if (enabled === true && !activeMap.has(d.code)) {
+      activeMap.set(d.code, d);
+      steps.push(`Override ${d.code} enabled by config -> +1 dimension`);
+    }
+  }
+
+  // Step 4: --dimension adds
+  if (flags.dimension && flags.dimension.length > 0) {
+    let added = 0;
+    for (const id of flags.dimension) {
+      const d = resolveDimension(id);
+      if (d && !activeMap.has(d.code)) {
+        activeMap.set(d.code, d);
+        added++;
+        steps.push(`--dimension ${d.code} added -> +1 dimension`);
+      }
+    }
+  }
+
+  // Step 5: --skip removes
+  if (flags.skip && flags.skip.length > 0) {
+    for (const id of flags.skip) {
+      const d = resolveDimension(id);
+      if (d && activeMap.has(d.code)) {
+        activeMap.delete(d.code);
+        steps.push(`--skip ${d.code} removed -> -1 dimension`);
+      }
+    }
+  }
+
+  steps.push(`Final: ${activeMap.size} active dimensions`);
+  return steps;
+}
+
+// ---------------------------------------------------------------------------
+// Exports
+// ---------------------------------------------------------------------------
+module.exports = {
+  DIMENSIONS,
+  CATEGORIES,
+  PRESETS,
+  getDimensionById,
+  getDimensionBySlug,
+  getDimensionsByCategory,
+  getActivePresetCategories,
+  resolveDimensions,
+  explainResolution,
+};