monomind 1.10.47 → 1.10.55

package/.claude/helpers/router.cjs

@@ -1,559 +0,0 @@
-#!/usr/bin/env node
-/**
- * Monomind Agent Router
- * Routes tasks to optimal agents based on learned patterns.
- * Also does keyword-matching against skill-registry.json (dev skills)
- * and extras-registry.json (non-dev specialist agents).
- */
-
-const path = require('path');
-const fs = require('fs');
-
-const AGENT_CAPABILITIES = {
-  coder: ['code-generation', 'refactoring', 'debugging', 'implementation'],
-  tester: ['unit-testing', 'integration-testing', 'coverage', 'test-generation'],
-  reviewer: ['code-review', 'security-audit', 'quality-check', 'best-practices'],
-  researcher: ['web-search', 'documentation', 'analysis', 'summarization'],
-  architect: ['system-design', 'architecture', 'patterns', 'scalability'],
-  'backend-dev': ['api', 'database', 'server', 'authentication'],
-  'frontend-dev': ['ui', 'react', 'css', 'components'],
-  devops: ['ci-cd', 'docker', 'deployment', 'infrastructure'],
-};
-
-// Maps generic role → specific specialized agents available in the system
-const SPECIFIC_AGENTS_MAP = {
-  coder: [
-    { slug: 'sparc-coder',       label: 'sparc-coder',        note: 'TDD + SPARC methodology' },
-    { slug: 'backend-dev',       label: 'backend-dev',         note: 'API, DB, server-side' },
-    { slug: 'frontend-dev',      label: 'Frontend Developer',  note: 'React/Vue/CSS' },
-    { slug: 'mobile-dev',        label: 'mobile-dev',          note: 'React Native iOS/Android' },
-    { slug: 'ml-developer',      label: 'ml-developer',        note: 'ML model dev & training' },
-  ],
-  tester: [
-    { slug: 'tdd-london-swarm',      label: 'tdd-london-swarm',       note: 'Mock-driven TDD' },
-    { slug: 'API Tester',            label: 'API Tester',              note: 'API validation & performance' },
-    { slug: 'production-validator',  label: 'production-validator',    note: 'Deployment readiness' },
-    { slug: 'Evidence Collector',    label: 'Evidence Collector',      note: 'Screenshot-backed QA' },
-    { slug: 'agent-browser-testing', label: 'agent-browser-testing',   note: 'UI/browser automation testing' },
-  ],
-  reviewer: [
-    { slug: 'Code Reviewer',              label: 'Code Reviewer',            note: 'Correctness, security, perf' },
-    { slug: 'code-analyzer',              label: 'code-analyzer',            note: 'Quality metrics & smells' },
-    { slug: 'feature-dev:code-reviewer',  label: 'feature-dev:code-reviewer',note: 'Bug & logic error detection' },
-    { slug: 'Reality Checker',            label: 'Reality Checker',          note: 'Evidence-based certification' },
-    { slug: 'Accessibility Auditor',      label: 'Accessibility Auditor',    note: 'WCAG & assistive tech' },
-  ],
-  researcher: [
-    { slug: 'sparc:researcher',  label: 'sparc:researcher',  note: 'Parallel web search + memory' },
-    { slug: 'Explore',           label: 'Explore',            note: 'Fast codebase exploration' },
-    { slug: 'Trend Researcher',  label: 'Trend Researcher',   note: 'Market intelligence' },
-    { slug: 'UX Researcher',     label: 'UX Researcher',      note: 'User behaviour & usability' },
-  ],
-  architect: [
-    { slug: 'system-architect',   label: 'system-architect',   note: 'High-level system design' },
-    { slug: 'Software Architect', label: 'Software Architect',  note: 'DDD, patterns, decisions' },
-    { slug: 'Backend Architect',  label: 'Backend Architect',   note: 'Scalable server-side design' },
-    { slug: 'Plan',               label: 'Plan',                note: 'Implementation strategy' },
-  ],
-  'backend-dev': [
-    { slug: 'backend-dev',        label: 'backend-dev',         note: 'API & server patterns' },
-    { slug: 'Database Optimizer', label: 'Database Optimizer',  note: 'Schema, indexes, query tuning' },
-    { slug: 'Data Engineer',      label: 'Data Engineer',       note: 'Pipelines, ETL, lakehouse' },
-    { slug: 'Security Engineer',  label: 'Security Engineer',   note: 'Threat modelling, secure code' },
-  ],
-  'frontend-dev': [
-    { slug: 'Frontend Developer', label: 'Frontend Developer',  note: 'React/Vue/Angular' },
-    { slug: 'UI Designer',        label: 'UI Designer',          note: 'Design systems & components' },
-    { slug: 'UX Architect',       label: 'UX Architect',         note: 'CSS systems & interaction' },
-    { slug: 'mobile-dev',         label: 'mobile-dev',           note: 'Cross-platform mobile' },
-  ],
-  devops: [
-    { slug: 'DevOps Automator',   label: 'DevOps Automator',    note: 'CI/CD, infra automation' },
-    { slug: 'SRE',                label: 'SRE',                  note: 'SLOs, reliability, on-call' },
-    { slug: 'cicd-engineer',      label: 'cicd-engineer',        note: 'GitHub Actions pipelines' },
-    { slug: 'Incident Response Commander', label: 'Incident Response Commander', note: 'Prod incident mgmt' },
-  ],
-};
-
-const TASK_PATTERNS = {
-  '\\bimplement\\b|\\bcreate\\b|\\bbuild\\b|\\badd\\b|\\bwrite\\s+code\\b': 'coder',
-  '\\btest\\b|\\bspec\\b|\\bcoverage\\b|unit test|\\bintegration\\b': 'tester',
-  '\\breview\\b|\\baudit\\b|\\bcheck\\b|\\bvalidate\\b|\\bsecurity\\b': 'reviewer',
-  '\\bresearch\\b|\\bfind\\b|\\bsearch\\b|\\bdocumentation\\b|\\bexplore\\b|\\bexplain\\b|\\bunderstand\\b|\\bhow does\\b|\\bhow do\\b|\\bwhat is\\b': 'researcher',
-  '\\bdesign\\b|\\barchitect\\b|\\bstructure\\b|\\bplan\\b': 'architect',
-  '\\bapi\\b|\\bendpoint\\b|\\bserver\\b|\\bbackend\\b|\\bdatabase\\b': 'backend-dev',
-  '\\bui\\b|\\bfrontend\\b|\\bcomponent\\b|\\breact\\b|\\bcss\\b|\\bstyle\\b': 'frontend-dev',
-  '\\bdeploy\\b|\\bdocker\\b|\\bci\\b|\\bcd\\b|\\bpipeline\\b|\\binfrastructure\\b': 'devops',
-};
-
-// Non-dev domain keywords — if matched, skip dev routing and go to extras
-const NON_DEV_PATTERNS = [
-  'marketing', 'campaign', 'social media', 'tiktok', 'instagram', 'twitter', 'linkedin',
-  'seo', 'content creation', 'viral', 'growth hacking', 'brand', 'influencer', 'ecommerce',
-  'sales', 'crm', 'sales pipeline', 'leads', 'prospects', 'quota', 'outbound',
-  'paid media', 'ppc', 'google ads', 'facebook ads', 'programmatic', 'display ads',
-  'product management', 'product roadmap', 'sprint', 'backlog', 'user story', 'customer feedback',
-  'project management', 'milestone', 'stakeholder', 'jira', 'agile', 'scrum',
-  'ux research', 'user research', 'usability', 'wireframe', 'prototype', 'figma',
-  'ui design', 'visual design', 'illustration', 'branding',
-  'academic', 'anthropology', 'history', 'geography', 'psychology', 'narrative theory',
-  'blockchain', 'salesforce', 'healthcare', 'compliance', 'supply chain',
-  'recruitment', 'hiring', 'human resources', 'finance tracking', 'invoice', 'executive summary',
-  'customer support', 'helpdesk', 'podcast', 'video editing', 'short video',
-];
-
-// ─── Skill registry (dev skills) ────────────────────────────────────────────
-let _skillRegistry = null;
-function loadSkillRegistry() {
-  if (_skillRegistry) return _skillRegistry;
-  try {
-    _skillRegistry = JSON.parse(fs.readFileSync(path.join(__dirname, 'skill-registry.json'), 'utf8'));
-  } catch (e) { _skillRegistry = { skills: [] }; }
-  return _skillRegistry;
-}
-
-// ─── Extras registry (non-dev agents) ───────────────────────────────────────
-let _extrasRegistry = null;
-function loadExtrasRegistry() {
-  if (_extrasRegistry) return _extrasRegistry;
-  try {
-    _extrasRegistry = JSON.parse(fs.readFileSync(path.join(__dirname, 'extras-registry.json'), 'utf8'));
-  } catch (e) { _extrasRegistry = { extras: [] }; }
-  return _extrasRegistry;
-}
-
-// ─── Scoring helpers ─────────────────────────────────────────────────────────
-function scoreEntry(keywords, taskLower) {
-  if (!Array.isArray(keywords)) return 0;
-  let score = 0;
-  for (const kw of keywords) {
-    if (typeof kw === 'string' && taskLower.includes(kw.toLowerCase())) score++;
-  }
-  return score;
-}
-
-function matchSkills(task, topN = 5) {
-  if (typeof task !== 'string') return [];
-  const registry = loadSkillRegistry();
-  const taskLower = task.toLowerCase();
-  return registry.skills
-    .map(s => ({ skill: s.skill, invoke: s.invoke, description: s.description, category: s.category, score: scoreEntry(s.keywords, taskLower) }))
-    .filter(s => s.score > 0)
-    .sort((a, b) => b.score - a.score)
-    .slice(0, topN);
-}
-
-// Categories that are almost never relevant in a software project context.
-// We exclude these from extras suggestions UNLESS the prompt has explicit
-// keywords that opt them back in. Default-conservative: in a code repo,
-// an "Anthropologist" for "community docs" or "Unity Architect" for a
-// "community module" is noise, not signal.
-const _NON_DEV_EXTRA_CATEGORIES = new Set([
-  'academic',         // Anthropologist, Geographer, Historian, Narratologist, Psychologist
-  'game-development', // Unity/Unreal/Godot/Blender — irrelevant for non-game codebases
-  'marketing',
-  'sales',
-  'paid-media',
-  'design',           // brand/UI/UX specialists — irrelevant for code reviews
-]);
-
-// Per-category opt-in keywords. If the prompt mentions one of these, that
-// category is allowed back into extras.
-const _CATEGORY_OPTIN_KEYWORDS = {
-  'academic':         /\b(anthropolog|sociolog|cultural|ritual|kinship|ethnograph|histor(?:y|ical|ian)|geograph|narratolog|psycholog|belief\s*system)\b/i,
-  'game-development': /\b(unity|unreal|godot|blender|game\s*(?:design|dev|engine)|gameplay|shader\s*graph|level\s*design|ugc|roblox)\b/i,
-  'marketing':        /\b(marketing|advertis|seo|tiktok|instagram|linkedin\s*content|brand|campaign|growth\s*hack|copywrit|blog\s*post|press\s*release|launch\s*strategy|pitch|investor)\b/i,
-  'sales':            /\b(sales|prospect|outbound|pipeline|crm|cold\s*email|deal|account\s*strategy|quota|forecast|qbr)\b/i,
-  'paid-media':       /\b(paid\s*media|ads?|google\s*ads|meta\s*ads|programmatic|ppc|cpa|conversion\s*tracking)\b/i,
-  'design':           /\b(figma|brand\s*identity|design\s*system|ux\s*research|persona|usability|wireframe|prototype|color\s*palette|typography|moodboard)\b/i,
-};
-
-function _allowedExtraCategories(taskLower) {
-  const allowed = new Set();
-  for (const [cat, re] of Object.entries(_CATEGORY_OPTIN_KEYWORDS)) {
-    if (re.test(taskLower)) allowed.add(cat);
-  }
-  return allowed;
-}
-
-function matchExtras(task, topN = 8) {
-  if (typeof task !== 'string') return [];
-  const registry = loadExtrasRegistry();
-  const taskLower = task.toLowerCase();
-  const optedIn = _allowedExtraCategories(taskLower);
-  return registry.extras
-    .filter(e => !_NON_DEV_EXTRA_CATEGORIES.has(e.category) || optedIn.has(e.category))
-    .map(e => ({ slug: e.slug, name: e.name, description: e.description, category: e.category, filePath: e.filePath, score: scoreEntry(e.keywords, taskLower) }))
-    .filter(e => e.score > 0)
-    .sort((a, b) => b.score - a.score)
-    .slice(0, topN);
-}
-
-// Pre-compiled word-boundary regexes for non-dev patterns (built once at load time)
-const _nonDevRegexes = NON_DEV_PATTERNS.map(function(kw) {
-  const escaped = kw.replace(/[.*+?^${}()|[\]\\]/g, '\\$&').replace(/\s+/g, '\\s+');
-  return new RegExp('\\b' + escaped + '\\b', 'i');
-});
-
-function isNonDevTask(taskLower) {
-  for (const re of _nonDevRegexes) {
-    if (re.test(taskLower)) return true;
-  }
-  return false;
-}
-
-// ─── Two-Stage Keyword Router (for non-dev and ambiguous tasks) ──────────────
-// Stage 1: pick best category by scoring task tokens against category keywords
-// Stage 2: pick best agent within that category by keyword overlap
-// Pure in-process — no external API calls, no API key required.
-
-function buildCategoryList() {
-  const registry = loadExtrasRegistry();
-  const cats = {};
-  for (const e of registry.extras) {
-    if (!cats[e.category]) cats[e.category] = [];
-    cats[e.category].push(e.name);
-  }
-  return Object.entries(cats).map(([name, agents]) => ({
-    name,
-    count: agents.length,
-    examples: agents.slice(0, 4).join(', '),
-  }));
-}
-
-function getAgentsInCategory(category) {
-  const registry = loadExtrasRegistry();
-  return registry.extras
-    .filter(e => e.category === category)
-    .map(e => ({ slug: e.slug, name: e.name, description: (e.description || '').slice(0, 120) }));
-}
-
-// Score a task string against an agent entry using keyword overlap
-function scoreAgainstEntry(taskTokens, entry) {
-  let score = 0;
-  const nameTokens = (entry.name || '').toLowerCase().split(/\W+/).filter(Boolean);
-  const descTokens = (entry.description || '').toLowerCase().split(/\W+/).filter(Boolean);
-  const kwTokens = Array.isArray(entry.keywords) ? entry.keywords.map(k => k.toLowerCase()) : [];
-  const allTokens = new Set([...nameTokens, ...descTokens, ...kwTokens]);
-  for (const t of taskTokens) {
-    if (allTokens.has(t)) score += 2;
-    else for (const s of allTokens) { if (s.includes(t) || t.includes(s)) { score += 1; break; } }
-  }
-  return score;
-}
-
-// Category-level score: sum agent scores within the category
-function scoreCategoryForTask(taskTokens, category, registry) {
-  const agents = registry.extras.filter(e => e.category === category);
-  return agents.reduce((sum, e) => sum + scoreAgainstEntry(taskTokens, e), 0);
-}
-
-async function routeTaskLLM(task) {
-  const registry = loadExtrasRegistry();
-  if (!registry.extras.length) return null;
-
-  const taskLower = task.toLowerCase();
-  const STOP = new Set(['the','and','for','are','but','not','you','all','can','had','her','was','one','our','out','day','get','has','him','his','how','its','may','new','now','old','see','two','way','who','did','let','put','say','she','too','use','will','with','that','this','from','they','been','have','make','then','than','when','what','does','into','your']);
-  const taskTokens = taskLower.split(/\W+/).filter(w => w.length > 3 && !STOP.has(w));
-
-  // Stage 1: pick best category by aggregate keyword score
-  const categories = [...new Set(registry.extras.map(e => e.category))];
-  const catScores = categories.map(cat => ({
-    cat,
-    score: scoreCategoryForTask(taskTokens, cat, registry),
-  })).sort((a, b) => b.score - a.score);
-
-  const bestCat = catScores[0];
-  if (!bestCat || bestCat.score === 0) return null;
-
-  // Stage 2: pick best agent within the category
-  const candidates = registry.extras.filter(e => e.category === bestCat.cat);
-  const scored = candidates.map(e => ({ e, score: scoreAgainstEntry(taskTokens, e) }))
-    .sort((a, b) => b.score - a.score);
-
-  const top = scored[0];
-  if (!top || top.score === 0) return null;
-
-  const allInCat = candidates.map(e => ({ slug: e.slug, label: e.name, note: e.category }));
-  return {
-    agent: top.e.name,
-    agentSlug: top.e.slug,
-    confidence: Math.min(0.85, 0.5 + top.score * 0.05),
-    reason: `Keyword 2-stage: ${bestCat.cat} → ${top.e.name}`,
-    category: bestCat.cat,
-    allInCategory: allInCat,
-  };
-}
-
-/**
- * Route multiple subtasks for swarm agent selection.
- * Each subtask description gets its own 2-stage LLM routing.
- * Returns array of { subtask, agent, agentSlug, confidence, reason }.
- */
-async function routeSwarmAgents(subtasks) {
-  if (!Array.isArray(subtasks) || !subtasks.length) return [];
-  const results = await Promise.all(subtasks.map(async (sub) => {
-    const desc = typeof sub === 'string' ? sub : sub.description || sub.task || '';
-    if (!desc) return { subtask: desc, agent: 'coder', agentSlug: 'coder', confidence: 0.5, reason: 'empty subtask' };
-    const llm = await routeTaskLLM(desc);
-    if (llm) return { subtask: desc, agent: llm.agent, agentSlug: llm.agentSlug, confidence: llm.confidence, reason: llm.reason };
-    const kw = routeTask(desc);
-    return { subtask: desc, agent: kw.agent, agentSlug: kw.agentSlug, confidence: kw.confidence, reason: kw.reason };
-  }));
-  return results;
-}
-
-// ─── RouteLayer bridge (GAP-002) ─────────────────────────────────────────────
-// Cache a Promise so concurrent callers all await the same load operation.
-var _routeLayerPromise = null;
-
-async function tryLoadRouteLayer() {
-  if (!_routeLayerPromise) {
-    _routeLayerPromise = (async function() {
-      try {
-        var routingModule = await import('@monomind/routing');
-        if (routingModule && routingModule.RouteLayer && routingModule.ALL_ROUTES) {
-          return new routingModule.RouteLayer({ routes: routingModule.ALL_ROUTES });
-        }
-      } catch (e) { /* @monomind/routing not compiled — keyword fallback will be used */ }
-      return null;
-    })();
-  }
-  return _routeLayerPromise;
-}
-
-/**
- * Async variant — tries LLM 2-stage routing for non-dev tasks,
- * RouteLayer semantic routing for dev tasks, falls back to keywords.
- */
-async function routeTaskSemantic(task) {
-  if (typeof task !== 'string' || !task) return routeTask(task);
-  const taskLower = task.toLowerCase();
-
-  // For non-dev tasks or ambiguous defaults, try LLM 2-stage routing first
-  if (isNonDevTask(taskLower)) {
-    const llmResult = await routeTaskLLM(task);
-    if (llmResult) {
-      const extrasMatches = matchExtras(task);
-      return {
-        agent: llmResult.agent,
-        agentSlug: llmResult.agentSlug,
-        confidence: llmResult.confidence,
-        reason: llmResult.reason,
-        skillMatches: [],
-        extrasMatches,
-        specificAgents: llmResult.allInCategory.slice(0, 5),
-        llmRouting: true,
-      };
-    }
-    // LLM failed — fall through to keyword-based extras matching
-  }
-
-  // Dev tasks: try RouteLayer semantic routing
-  const rl = await tryLoadRouteLayer();
-  if (rl && rl.route) {
-    try {
-      const semantic = await rl.route(task);
-      if (semantic && semantic.agentSlug && semantic.confidence > 0.6) {
-        const mapEntry = SPECIFIC_AGENTS_MAP[semantic.agentSlug];
-        const isDevDomain = !!(mapEntry && mapEntry.length > 0);
-        const extrasForTask = isDevDomain ? [] : matchExtras(task);
-        const semanticSpecificAgents = (mapEntry && mapEntry.length > 0)
-          ? mapEntry
-          : extrasForTask.slice(0, 5).map(e => ({ slug: e.slug, label: e.name, note: e.category }));
-        return {
-          agent: semantic.agentSlug,
-          agentSlug: semantic.agentSlug,
-          confidence: semantic.confidence,
-          reason: 'RouteLayer semantic (' + (semantic.method || 'semantic') + '): ' + semantic.routeName,
-          skillMatches: matchSkills(task),
-          extrasMatches: extrasForTask,
-          specificAgents: semanticSpecificAgents,
-          semanticRouting: true,
-        };
-      }
-    } catch (e) { /* fall through to keyword */ }
-  }
-
-  // Default keyword fallback — also try LLM if no dev pattern matched
-  const keywordResult = routeTask(task);
-  if (keywordResult.confidence <= 0.5) {
-    const llmResult = await routeTaskLLM(task);
-    if (llmResult) {
-      return {
-        agent: llmResult.agent,
-        agentSlug: llmResult.agentSlug,
-        confidence: llmResult.confidence,
-        reason: llmResult.reason,
-        skillMatches: keywordResult.skillMatches,
-        extrasMatches: matchExtras(task),
-        specificAgents: llmResult.allInCategory.slice(0, 5),
-        llmRouting: true,
-      };
-    }
-  }
-  return keywordResult;
-}
-
-// ─── Feedback weight loader ───────────────────────────────────────────────────
-// Reads .monomind/routing-feedback.jsonl to compute per-agent success-rate
-// weights. Weights are memoized with a 60-second TTL.
-//
-// Signal source: routing-feedback.jsonl entries with a non-null intelligenceFeedback
-// boolean. Derived at session-end from intelligence-outcomes.jsonl (30-minute window,
-// majority-vote). Requires MIN_FEEDBACK_SAMPLES entries per agent to activate.
-//
-// To verify: copy .monomind/test-fixtures/routing-feedback-seeded.jsonl to
-// .monomind/routing-feedback.jsonl, then call loadFeedbackWeights() in a REPL.
-
-var _feedbackWeightsCache = null;
-var _feedbackWeightsCacheTime = 0;
-var FEEDBACK_CACHE_TTL_MS = 60000;
-var MIN_FEEDBACK_SAMPLES = 10;
-
-function loadFeedbackWeights() {
-  var now = Date.now();
-  if (_feedbackWeightsCache && now - _feedbackWeightsCacheTime < FEEDBACK_CACHE_TTL_MS) {
-    return _feedbackWeightsCache;
-  }
-  var feedbackPath = path.join(process.cwd(), '.monomind', 'routing-feedback.jsonl');
-  if (!fs.existsSync(feedbackPath)) {
-    _feedbackWeightsCache = new Map();
-    _feedbackWeightsCacheTime = now;
-    return _feedbackWeightsCache;
-  }
-  var lines;
-  try {
-    lines = fs.readFileSync(feedbackPath, 'utf8').trim().split('\n').filter(Boolean);
-  } catch {
-    _feedbackWeightsCache = new Map();
-    _feedbackWeightsCacheTime = now;
-    return _feedbackWeightsCache;
-  }
-  var agentStats = {};
-  for (var i = 0; i < lines.length; i++) {
-    try {
-      var entry = JSON.parse(lines[i]);
-      var agent = entry.suggestedAgent || entry.agent;
-      if (!agent) continue;
-      if (!agentStats[agent]) agentStats[agent] = { total: 0, successTotal: 0, withSignal: 0 };
-      agentStats[agent].total++;
-      // Only count entries where intelligenceFeedback is non-null as having signal
-      var fb = entry.intelligenceFeedback;
-      if (fb !== null && fb !== undefined) {
-        agentStats[agent].withSignal++;
-        var success = fb === true || (fb && fb.success === true);
-        if (success) agentStats[agent].successTotal++;
-      }
-    } catch { /* skip malformed lines */ }
-  }
-  var weights = new Map();
-  for (var agentName in agentStats) {
-    var stats = agentStats[agentName];
-    // Only activate weight when we have enough non-null signal entries
-    if (stats.withSignal >= MIN_FEEDBACK_SAMPLES) {
-      var rate = stats.successTotal / stats.withSignal;
-      // Clamp to [0.5, 1.5]: bad agents get penalized, good agents boosted slightly
-      weights.set(agentName, Math.max(0.5, Math.min(1.5, rate * 2)));
-    }
-  }
-  _feedbackWeightsCache = weights;
-  _feedbackWeightsCacheTime = now;
-  return weights;
-}
-
-// ─── Main routing ─────────────────────────────────────────────────────────────
-function routeTask(task) {
-  if (typeof task !== 'string' || !task) {
-    return { agent: 'coder', agentSlug: 'coder', confidence: 0, reason: 'Empty task', skillMatches: [], extrasMatches: [], specificAgents: SPECIFIC_AGENTS_MAP['coder'] || [] };
-  }
-  const taskLower = task.toLowerCase();
-
-  // Check non-dev first — resolve to the top-matched specialist agent name (never "extras")
-  if (isNonDevTask(taskLower)) {
-    const extrasMatches = matchExtras(task);
-    const top = extrasMatches[0];
-    const agentName = top ? top.name : 'Specialist Agent';
-    const agentSlug = top ? top.slug : 'specialist-agent';
-    return {
-      agent: agentName,
-      agentSlug: agentSlug,
-      confidence: top ? 0.85 : 0.5,
-      reason: 'Domain: ' + (top ? top.category : 'non-dev') + ' | /specialagent',
-      skillMatches: [],
-      extrasMatches,
-      specificAgents: extrasMatches.slice(0, 5).map(e => ({ slug: e.slug, label: e.name, note: e.category })),
-    };
-  }
-
-  var feedbackWeights = loadFeedbackWeights();
-
-  // Dev task pattern matching
-  for (const [pattern, agent] of Object.entries(TASK_PATTERNS)) {
-    const regex = new RegExp(pattern, 'i');
-    if (regex.test(taskLower)) {
-      // Apply feedback weight to confidence (defaults to 1.0 = no change)
-      var baseConfidence = 0.8;
-      var weight = feedbackWeights.get(agent) || 1.0;
-      return {
-        agent,
-        agentSlug: agent,
-        confidence: Math.min(1.0, baseConfidence * weight),
-        reason: `Matched pattern: ${pattern}`,
-        skillMatches: matchSkills(task),
-        extrasMatches: [],
-        specificAgents: SPECIFIC_AGENTS_MAP[agent] || [],
-      };
-    }
-  }
-
-  // Default — low confidence, show both skill and extras suggestions
-  var defaultAgent = 'coder';
-  var defaultWeight = feedbackWeights.get(defaultAgent) || 1.0;
-  return {
-    agent: defaultAgent,
-    agentSlug: defaultAgent,
-    confidence: Math.min(1.0, 0.5 * defaultWeight),
-    reason: 'Default routing - no specific pattern matched',
-    skillMatches: matchSkills(task),
-    extrasMatches: matchExtras(task),
-    specificAgents: SPECIFIC_AGENTS_MAP[defaultAgent] || [],
-  };
-}
-
-/**
- * Load the full text of an extras agent by slug.
- * Used when Claude picks an agent to activate.
- */
-function loadExtrasAgent(slug) {
-  if (typeof slug !== 'string' || !slug) return null;
-  const registry = loadExtrasRegistry();
-  const slugLower = slug.toLowerCase();
-  const entry = registry.extras.find(e =>
-    e.slug === slug || (typeof e.name === 'string' && e.name.toLowerCase() === slugLower)
-  );
-  if (!entry) return null;
-  try {
-    const resolvedPath = path.isAbsolute(entry.filePath)
-      ? entry.filePath
-      : path.resolve(__dirname, entry.filePath);
-    return { ...entry, content: fs.readFileSync(resolvedPath, 'utf8') };
-  } catch (e) { return null; }
-}
-
-module.exports = { routeTask, routeTaskSemantic, routeTaskLLM, routeSwarmAgents, matchSkills, matchExtras, loadExtrasAgent, loadExtrasRegistry, loadSkillRegistry, buildCategoryList, getAgentsInCategory, AGENT_CAPABILITIES, TASK_PATTERNS };
-
-// CLI
-if (require.main === module) {
-  const args = process.argv.slice(2);
-  if (args[0] === '--load-agent') {
-    const agent = loadExtrasAgent(args.slice(1).join(' '));
-    if (agent) { console.log(agent.content); }
-    else { console.error('Agent not found'); process.exit(1); }
-  } else if (args.length) {
-    const result = routeTask(args.join(' '));
-    console.log(JSON.stringify(result, null, 2));
-  } else {
-    console.log('Usage: router.cjs <task>  OR  router.cjs --load-agent <slug>');
-  }
-}

package/.claude/helpers/session.cjs

@@ -1,126 +0,0 @@
-#!/usr/bin/env node
-/**
- * Monomind Cross-Platform Session Manager
- * Works on Windows, macOS, and Linux
- */
-
-const fs = require('fs');
-const path = require('path');
-const os = require('os');
-
-const platform = os.platform();
-const homeDir = os.homedir();
-
-function getDataDir() {
-  const baseDir = process.env.CLAUDE_PROJECT_DIR || process.cwd();
-  const localDir = path.join(baseDir, '.monomind', 'sessions');
-  if (fs.existsSync(path.dirname(localDir))) {
-    return localDir;
-  }
-
-  switch (platform) {
-    case 'win32':
-      return path.join(process.env.APPDATA || homeDir, 'monomind', 'sessions');
-    case 'darwin':
-      return path.join(homeDir, 'Library', 'Application Support', 'monomind', 'sessions');
-    default:
-      return path.join(homeDir, '.monomind', 'sessions');
-  }
-}
-
-const SESSION_DIR = getDataDir();
-const SESSION_FILE = path.join(SESSION_DIR, 'current.json');
-
-function ensureDir(dir) {
-  if (!fs.existsSync(dir)) {
-    fs.mkdirSync(dir, { recursive: true });
-  }
-}
-
-const commands = {
-  start: () => {
-    ensureDir(SESSION_DIR);
-    const sessionId = `session-${Date.now()}`;
-    const session = {
-      id: sessionId,
-      startedAt: new Date().toISOString(),
-      platform: platform,
-      cwd: process.cwd(),
-      context: {},
-      metrics: { edits: 0, commands: 0, tasks: 0, errors: 0 }
-    };
-    fs.writeFileSync(SESSION_FILE, JSON.stringify(session, null, 2));
-    console.log(`Session started: ${sessionId}`);
-    return session;
-  },
-
-  restore: () => {
-    if (!fs.existsSync(SESSION_FILE)) {
-      console.log('No session to restore');
-      return null;
-    }
-    const session = JSON.parse(fs.readFileSync(SESSION_FILE, 'utf-8'));
-    session.restoredAt = new Date().toISOString();
-    fs.writeFileSync(SESSION_FILE, JSON.stringify(session, null, 2));
-    console.log(`Session restored: ${session.id}`);
-    return session;
-  },
-
-  end: () => {
-    if (!fs.existsSync(SESSION_FILE)) {
-      console.log('No active session');
-      return null;
-    }
-    const session = JSON.parse(fs.readFileSync(SESSION_FILE, 'utf-8'));
-    session.endedAt = new Date().toISOString();
-    session.duration = Date.now() - new Date(session.startedAt).getTime();
-
-    const archivePath = path.join(SESSION_DIR, `${session.id}.json`);
-    fs.writeFileSync(archivePath, JSON.stringify(session, null, 2));
-    fs.unlinkSync(SESSION_FILE);
-
-    console.log(`Session ended: ${session.id}`);
-    console.log(`Duration: ${Math.round(session.duration / 1000 / 60)} minutes`);
-    return session;
-  },
-
-  status: () => {
-    if (!fs.existsSync(SESSION_FILE)) {
-      console.log('No active session');
-      return null;
-    }
-    const session = JSON.parse(fs.readFileSync(SESSION_FILE, 'utf-8'));
-    const duration = Date.now() - new Date(session.startedAt).getTime();
-    console.log(`Session: ${session.id}`);
-    console.log(`Platform: ${session.platform}`);
-    console.log(`Started: ${session.startedAt}`);
-    console.log(`Duration: ${Math.round(duration / 1000 / 60)} minutes`);
-    return session;
-  },
-
-  metric: (name) => {
-    if (!fs.existsSync(SESSION_FILE)) {
-      return null;
-    }
-    const session = JSON.parse(fs.readFileSync(SESSION_FILE, 'utf-8'));
-    if (session.metrics[name] !== undefined) {
-      session.metrics[name]++;
-      fs.writeFileSync(SESSION_FILE, JSON.stringify(session, null, 2));
-    }
-    return session;
-  }
-};
-
-module.exports = commands;
-
-// CLI - only run when executed directly
-if (require.main === module) {
-  const [,, command, ...args] = process.argv;
-  if (command && commands[command]) {
-    commands[command](...args);
-  } else {
-    console.log('Usage: session.js <start|restore|end|status|metric>');
-    console.log(`Platform: ${platform}`);
-    console.log(`Data dir: ${SESSION_DIR}`);
-  }
-}