@defai.digital/ax-cli 4.4.18 → 5.0.1

package/dist/setup.js

@@ -17,161 +17,114 @@
 import chalk from 'chalk';
 import { select, confirm, input } from '@inquirer/prompts';
 import ora from 'ora';
-import { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';
-import { homedir } from 'os';
-import { join } from 'path';
+import { existsSync } from 'fs';
 import { execSync, spawnSync } from 'child_process';
+import { AX_CLI_PROVIDER } from '@defai.digital/ax-core';
+import { AX_CLI_CONFIG_FILE, deleteConfig, loadConfig, saveConfig, } from './config.js';
+const DEFAULT_LOCAL_BASE_URL = AX_CLI_PROVIDER.defaultBaseURL || 'http://localhost:11434/v1';
 // ═══════════════════════════════════════════════════════════════════
-// 2025 OFFLINE CODING LLM RANKINGS (Updated December 2025)
+// MODEL DATA - Derived from AX_CLI_PROVIDER (single source of truth)
 // ═══════════════════════════════════════════════════════════════════
-//
-// Tier 1: Qwen 3 (32B/72B) - 9.6/10 - BEST OVERALL (PRIMARY)
-//   - Multi-task consistency and stability still best
-//   - Best for large codebase analysis (especially 32B+)
-//   - Highest agentic tool integration (AX-CLI)
-//   - Best Claude Code alternative for offline
-//   → Recommended as PRIMARY model
-//
-// Tier 2: GLM-4.6 (9B/32B) - 9.4/10 - BEST REFACTOR + DOCS (MAJOR UPGRADE!)
-//   - GLM-4.6-Coder coding significantly upgraded, cross-language excellent
-//   - 9B performance rivals Qwen 14B / DeepSeek 16B
-//   - Superior for large-scale refactor + documentation generation
-//   - Beats DeepSeek Coder V2 on long context reasoning
-//   - Best bilingual (Chinese + English) understanding
-//   → Recommended as SECONDARY core model
-//
-// Tier 3: DeepSeek-Coder V2 (7B/16B) - 9.3/10 - BEST SPEED
-//   - #1 inference speed
-//   - Easiest edge deployment (Jetson/Mac)
-//   - Best small model efficiency
-//   - BUT: GLM-4.6 9B > DeepSeek 16B on large codebase reasoning
-//   → Best for: quick patches, linting, small refactors
-//
-// Tier 4: Codestral/Mistral - 8.4/10 - C++/RUST NICHE
-//   - Niche advantage in C++/Rust
-//   - Overall capabilities superseded by Qwen/GLM/DeepSeek
-//
-// Tier 5: Llama 3.1/CodeLlama - 8.1/10 - FALLBACK
-//   - Wide ecosystem, best framework support
-//   - Stable and robust
-//   - No longer leading in capability
-// ═══════════════════════════════════════════════════════════════════
-// TIER 1: Qwen 3 - Best overall offline coding model (9.6/10)
-// → PRIMARY model for most coding tasks
-const LOCAL_QWEN_MODELS = [
-    { id: 'qwen3:72b', name: 'Qwen 3 72B', description: 'PRIMARY: Most capable, 128K context' },
-    { id: 'qwen3:32b', name: 'Qwen 3 32B', description: 'PRIMARY: Best for large codebase analysis' },
-    { id: 'qwen3:14b', name: 'Qwen 3 14B', description: 'PRIMARY: Balanced performance (recommended)' },
-    { id: 'qwen3:8b', name: 'Qwen 3 8B', description: 'PRIMARY: Efficient, great for most tasks' },
-    { id: 'qwen2.5-coder:32b', name: 'Qwen2.5-Coder 32B', description: 'Excellent coding specialist' },
-    { id: 'qwen2.5-coder:14b', name: 'Qwen2.5-Coder 14B', description: 'Balanced coding model' },
-];
-// TIER 2: GLM-4.6 - Best for refactor + docs (9.4/10) - MAJOR UPGRADE!
-// → SECONDARY core model, excellent for architecture refactoring
-// Note: For cloud GLM with web search/vision, use ax-glm
-const LOCAL_GLM_MODELS = [
-    { id: 'glm-4.6:32b', name: 'GLM-4.6 32B', description: 'REFACTOR: Large-scale refactor + multi-file editing' },
-    { id: 'glm-4.6:9b', name: 'GLM-4.6 9B', description: 'REFACTOR: Rivals Qwen 14B, excellent long context' },
-    { id: 'codegeex4', name: 'CodeGeeX4', description: 'DOCS: Best for documentation generation' },
-    { id: 'glm4:9b', name: 'GLM-4 9B', description: 'Bilingual code understanding' },
-];
-// TIER 3: DeepSeek-Coder V2 - Best speed (9.3/10)
-// → Best for quick iterations, patches, linting
-// Note: For cloud DeepSeek, a future ax-deepseek package will be available
-const LOCAL_DEEPSEEK_MODELS = [
-    { id: 'deepseek-coder-v2:16b', name: 'DeepSeek-Coder-V2 16B', description: 'SPEED: Fast iterations, patches' },
-    { id: 'deepseek-coder-v2:7b', name: 'DeepSeek-Coder-V2 7B', description: 'SPEED: 7B rivals 13B, edge-friendly' },
-    { id: 'deepseek-v3', name: 'DeepSeek V3', description: 'Latest general + coding model' },
-    { id: 'deepseek-coder:33b', name: 'DeepSeek-Coder 33B', description: 'Strong coding model' },
-    { id: 'deepseek-coder:6.7b', name: 'DeepSeek-Coder 6.7B', description: 'Efficient, low memory' },
-];
-// TIER 4: Codestral/Mistral - C++/Rust niche (8.4/10)
-const LOCAL_CODESTRAL_MODELS = [
-    { id: 'codestral:22b', name: 'Codestral 22B', description: 'C++/RUST: Systems programming niche' },
-    { id: 'mistral:7b', name: 'Mistral 7B', description: 'Good speed/accuracy balance' },
-    { id: 'mistral-nemo:12b', name: 'Mistral Nemo 12B', description: 'Compact but capable' },
-];
-// TIER 5: Llama - Best fallback/compatibility (8.1/10)
-const LOCAL_LLAMA_MODELS = [
-    { id: 'llama3.1:70b', name: 'Llama 3.1 70B', description: 'FALLBACK: Best framework compatibility' },
-    { id: 'llama3.1:8b', name: 'Llama 3.1 8B', description: 'FALLBACK: Fast, stable' },
-    { id: 'llama3.2:11b', name: 'Llama 3.2 11B', description: 'Vision support' },
-    { id: 'codellama:34b', name: 'Code Llama 34B', description: 'Code specialist' },
-    { id: 'codellama:7b', name: 'Code Llama 7B', description: 'Efficient code model' },
-];
-// All local models combined for offline setup (ordered by tier)
-const ALL_LOCAL_MODELS = [
-    // Tier 1: Qwen (PRIMARY - recommended for most coding tasks)
-    ...LOCAL_QWEN_MODELS.map(m => ({ ...m, name: `[T1-Qwen] ${m.name}` })),
-    // Tier 2: GLM-4.6 (REFACTOR - best for large-scale refactoring + docs)
-    ...LOCAL_GLM_MODELS.map(m => ({ ...m, name: `[T2-GLM] ${m.name}` })),
-    // Tier 3: DeepSeek (SPEED - best for quick iterations)
-    ...LOCAL_DEEPSEEK_MODELS.map(m => ({ ...m, name: `[T3-DeepSeek] ${m.name}` })),
-    // Tier 4: Codestral (C++/RUST - systems programming)
-    ...LOCAL_CODESTRAL_MODELS.map(m => ({ ...m, name: `[T4-Codestral] ${m.name}` })),
-    // Tier 5: Llama (FALLBACK - compatibility)
-    ...LOCAL_LLAMA_MODELS.map(m => ({ ...m, name: `[T5-Llama] ${m.name}` })),
-];
+/**
+ * Model tier configuration for categorization and display
+ * Single source of truth for tier metadata (colors, ratings, labels)
+ */
+const MODEL_TIERS = {
+    T1: { prefix: 'T1-Qwen', pattern: /qwen/i, rating: '9.6/10', label: 'PRIMARY', displayName: 'Qwen 3', description: 'Best overall, coding leader', color: chalk.green },
+    T2: { prefix: 'T2-GLM', pattern: /glm|codegeex|chatglm/i, rating: '9.4/10', label: 'REFACTOR', displayName: 'GLM', description: 'Large-scale refactor + docs', color: chalk.magenta, isNew: true },
+    T3: { prefix: 'T3-DeepSeek', pattern: /deepseek/i, rating: '9.3/10', label: 'SPEED', displayName: 'DeepSeek', description: 'Quick patches, linting', color: chalk.blue },
+    T4: { prefix: 'T4-Codestral', pattern: /codestral|mistral/i, rating: '8.4/10', label: 'C++/RUST', displayName: 'Codestral', description: 'Systems programming', color: chalk.cyan },
+    T5: { prefix: 'T5-Llama', pattern: /llama|codellama/i, rating: '8.1/10', label: 'FALLBACK', displayName: 'Llama', description: 'Best compatibility', color: chalk.gray },
+};
+/**
+ * Convert provider model config to setup ModelInfo format
+ */
+function providerModelsToModelInfo() {
+    return Object.entries(AX_CLI_PROVIDER.models).map(([id, config]) => ({
+        id,
+        name: config.name,
+        description: config.description,
+    }));
+}
+const PROVIDER_MODEL_INFOS = providerModelsToModelInfo();
+/**
+ * Get the tier for a model ID
+ */
+function getModelTier(modelId) {
+    const lower = modelId.toLowerCase();
+    for (const [tier, config] of Object.entries(MODEL_TIERS)) {
+        if (config.pattern.test(lower)) {
+            return { tier, label: config.prefix };
+        }
+    }
+    return { tier: 'T5', label: 'Other' };
+}
+/**
+ * Get models grouped by tier with prefixes
+ */
+function getModelsWithTierPrefix(models = PROVIDER_MODEL_INFOS) {
+    return models.map(m => {
+        const { label } = getModelTier(m.id);
+        return { ...m, name: label !== 'Other' ? `[${label}] ${m.name}` : m.name };
+    });
+}
+/**
+ * Get models for a specific tier
+ */
+function getModelsByTier(tier, models = PROVIDER_MODEL_INFOS) {
+    return models.filter(m => getModelTier(m.id).tier === tier);
+}
+// Derived model lists from single source of truth
+const ALL_LOCAL_MODELS = getModelsWithTierPrefix();
 // ═══════════════════════════════════════════════════════════════════
 // PROVIDER CONFIGURATION - LOCAL/OFFLINE ONLY
 // ═══════════════════════════════════════════════════════════════════
 const PROVIDER_INFO = {
-    name: 'Local/Offline (Ollama, LMStudio, vLLM)',
-    description: 'Run models locally - Qwen 3 recommended (best offline coding model)',
-    cliName: 'ax-cli',
-    package: '@ax-cli/cli',
-    defaultBaseURL: 'http://localhost:11434/v1',
-    defaultModel: 'qwen3:14b', // Tier 1: Best overall
-    apiKeyEnvVar: '', // No API key for local
+    name: AX_CLI_PROVIDER.displayName,
+    description: AX_CLI_PROVIDER.branding.description,
+    cliName: AX_CLI_PROVIDER.branding.cliName,
+    defaultBaseURL: DEFAULT_LOCAL_BASE_URL,
+    defaultModel: AX_CLI_PROVIDER.defaultModel,
     website: 'https://ollama.ai',
     models: ALL_LOCAL_MODELS,
 };
 // Well-known local server ports
 const LOCAL_SERVERS = [
-    { name: 'Ollama', url: 'http://localhost:11434/v1', port: 11434 },
+    { name: 'Ollama', url: DEFAULT_LOCAL_BASE_URL, port: 11434 },
     { name: 'LM Studio', url: 'http://localhost:1234/v1', port: 1234 },
     { name: 'vLLM', url: 'http://localhost:8000/v1', port: 8000 },
     { name: 'LocalAI', url: 'http://localhost:8080/v1', port: 8080 },
 ];
 // Config paths
-const AX_CLI_CONFIG_DIR = join(homedir(), '.ax-cli');
-const AX_CLI_CONFIG_FILE = join(AX_CLI_CONFIG_DIR, 'config.json');
 /**
- * Load ax-cli config
+ * Normalize base URLs to avoid trailing slash issues
  */
-function loadConfig() {
+function normalizeBaseURL(baseURL) {
     try {
-        if (existsSync(AX_CLI_CONFIG_FILE)) {
-            return JSON.parse(readFileSync(AX_CLI_CONFIG_FILE, 'utf-8'));
-        }
+        const parsed = new URL(baseURL);
+        parsed.pathname = parsed.pathname.replace(/\/+$/, '');
+        return parsed.toString().replace(/\/$/, '');
     }
     catch {
-        // Ignore errors
+        return baseURL.replace(/\/+$/, '');
     }
-    return {};
 }
 /**
- * Save ax-cli config
+ * Build models endpoint from a base URL
  */
-function saveConfig(config) {
-    if (!existsSync(AX_CLI_CONFIG_DIR)) {
-        mkdirSync(AX_CLI_CONFIG_DIR, { recursive: true });
-    }
-    writeFileSync(AX_CLI_CONFIG_FILE, JSON.stringify(config, null, 2), { mode: 0o600 });
+function buildModelsEndpoint(baseURL) {
+    return `${normalizeBaseURL(baseURL)}/models`;
 }
 /**
- * Delete ax-cli config (for --force flag)
+ * Run an async task with a spinner, ensuring cleanup even on failure
  */
-function deleteConfig() {
-    const { unlinkSync } = require('fs');
+async function withSpinner(text, task) {
+    const spinner = ora(text).start();
     try {
-        if (existsSync(AX_CLI_CONFIG_FILE)) {
-            unlinkSync(AX_CLI_CONFIG_FILE);
-        }
-        return true;
+        return await task();
     }
-    catch {
-        return false;
+    finally {
+        spinner.stop();
     }
 }
 /**
@@ -195,13 +148,13 @@ function getAutomatosXStatus() {
     }
 }
 /**
- * Install AutomatosX globally
+ * Execute a command with a timeout and inherited stdio
  */
-async function installAutomatosX() {
+function runCommand(command, timeoutMs) {
     try {
-        execSync('npm install -g @defai.digital/automatosx', {
+        execSync(command, {
             stdio: 'inherit',
-            timeout: 180000 // 3 minutes timeout
+            timeout: timeoutMs
         });
         return true;
     }
@@ -209,34 +162,30 @@ async function installAutomatosX() {
         return false;
     }
 }
+/**
+ * Install AutomatosX globally
+ */
+function installAutomatosX() {
+    return runCommand('npm install -g @defai.digital/automatosx', 180000);
+}
 /**
  * Run AutomatosX setup with force flag
  */
-async function runAutomatosXSetup() {
-    try {
-        execSync('ax setup -f', {
-            stdio: 'inherit',
-            timeout: 120000 // 2 minutes timeout
-        });
-        return true;
-    }
-    catch {
-        return false;
-    }
+function runAutomatosXSetup() {
+    return runCommand('ax setup -f', 120000);
 }
 /**
  * Fetch with timeout - reduces duplication in network calls
  */
 async function fetchWithTimeout(url, timeoutMs, options = {}) {
+    const controller = new AbortController();
+    const timeoutHandle = setTimeout(() => controller.abort(), timeoutMs);
     try {
-        const controller = new AbortController();
-        const timeoutHandle = setTimeout(() => controller.abort(), timeoutMs);
         const response = await fetch(url, {
             ...options,
             signal: controller.signal,
             headers: { 'Content-Type': 'application/json', ...options.headers },
         });
-        clearTimeout(timeoutHandle);
         if (!response.ok) {
             return { ok: false };
         }
@@ -246,19 +195,22 @@ async function fetchWithTimeout(url, timeoutMs, options = {}) {
     catch {
         return { ok: false };
     }
+    finally {
+        clearTimeout(timeoutHandle);
+    }
 }
 /**
  * Check if a local server is running on a given port
  */
 async function checkLocalServer(url) {
-    const result = await fetchWithTimeout(`${url}/models`, 2000);
+    const result = await fetchWithTimeout(buildModelsEndpoint(url), 2000);
     return result.ok;
 }
 /**
  * Fetch models from a local server
  */
 async function fetchLocalModels(baseURL) {
-    const result = await fetchWithTimeout(`${baseURL}/models`, 5000);
+    const result = await fetchWithTimeout(buildModelsEndpoint(baseURL), 5000);
     if (!result.ok || !result.data?.data) {
         return [];
     }
@@ -272,10 +224,14 @@ async function fetchLocalModels(baseURL) {
  * Detect running local servers
  */
 async function detectLocalServers() {
-    const results = await Promise.all(LOCAL_SERVERS.map(async (server) => ({
-        ...server,
-        available: await checkLocalServer(server.url),
-    })));
+    const results = await Promise.all(LOCAL_SERVERS.map(async (server) => {
+        const normalizedURL = normalizeBaseURL(server.url);
+        return {
+            ...server,
+            url: normalizedURL,
+            available: await checkLocalServer(normalizedURL),
+        };
+    }));
     return results;
 }
 /**
@@ -290,32 +246,15 @@ function validateURL(value) {
         return 'Please enter a valid URL';
     }
 }
-/**
- * Categorize model by tier (2025 rankings)
- */
-function categorizeModel(id) {
-    const lower = id.toLowerCase();
-    if (lower.includes('qwen'))
-        return { tier: 'T1', label: 'T1-Qwen' };
-    if (lower.includes('glm') || lower.includes('codegeex') || lower.includes('chatglm'))
-        return { tier: 'T2', label: 'T2-GLM' };
-    if (lower.includes('deepseek'))
-        return { tier: 'T3', label: 'T3-DeepSeek' };
-    if (lower.includes('codestral') || lower.includes('mistral'))
-        return { tier: 'T4', label: 'T4-Codestral' };
-    if (lower.includes('llama') || lower.includes('codellama'))
-        return { tier: 'T5', label: 'T5-Llama' };
-    return { tier: 'T6', label: 'Other' };
-}
 /**
  * Sort models by tier priority
  */
 function sortModelsByTier(models) {
-    const tierOrder = ['T1', 'T2', 'T3', 'T4', 'T5', 'T6'];
+    const tierOrder = ['T1', 'T2', 'T3', 'T4', 'T5'];
     return [...models].sort((a, b) => {
-        const catA = tierOrder.indexOf(categorizeModel(a.id).tier);
-        const catB = tierOrder.indexOf(categorizeModel(b.id).tier);
-        return catA - catB;
+        const tierA = tierOrder.indexOf(getModelTier(a.id).tier);
+        const tierB = tierOrder.indexOf(getModelTier(b.id).tier);
+        return (tierA === -1 ? 99 : tierA) - (tierB === -1 ? 99 : tierB);
     });
 }
 /**
@@ -323,7 +262,7 @@ function sortModelsByTier(models) {
  */
 function buildModelChoices(models, addTierPrefix = true) {
     const choices = models.map(m => {
-        const { label } = categorizeModel(m.id);
+        const { label } = getModelTier(m.id);
         const prefix = addTierPrefix && label !== 'Other' ? `[${label}] ` : '';
         return {
             name: `${prefix}${m.name} - ${m.description}`,
@@ -365,136 +304,148 @@ function printBoxHeader(title, width = 50) {
     console.log(chalk.cyan(`  └${'─'.repeat(width)}┘\n`));
 }
 /**
- * Run the setup wizard
+ * Get display limit for a tier (how many models to show)
  */
-export async function runSetup(options = {}) {
+function getTierDisplayLimit(tier) {
+    return tier === 'T4' ? 1 : tier === 'T5' ? 2 : 3;
+}
+/**
+ * Print model recommendations by tier
+ */
+function printModelRecommendations() {
+    console.log('  Recommended models by tier (Updated Dec 2025):\n');
+    for (const [tier, config] of Object.entries(MODEL_TIERS)) {
+        const models = getModelsByTier(tier).slice(0, getTierDisplayLimit(tier));
+        if (models.length === 0)
+            continue;
+        console.log(config.color.bold(`  ${config.prefix} (${config.rating}) - ${config.label}:`));
+        models.forEach(m => {
+            console.log(`    ${config.color('•')} ${m.id}: ${chalk.dim(m.description)}`);
+        });
+        console.log();
+    }
+}
+/**
+ * Print shared cloud provider guidance
+ */
+function printCloudProviderInfo() {
+    console.log(chalk.dim('  For cloud providers, use dedicated CLIs:'));
+    console.log(chalk.dim('  • GLM (Z.AI):     npm install -g @defai.digital/ax-glm && ax-glm setup'));
+    console.log(chalk.dim('  • Grok (xAI):     npm install -g @defai.digital/ax-grok && ax-grok setup'));
+    console.log(chalk.dim('  • DeepSeek:       (coming soon) @defai.digital/ax-deepseek'));
+    console.log();
+}
+/**
+ * Print welcome banner
+ */
+function printWelcomeBanner() {
     console.log(chalk.cyan('\n  ╔════════════════════════════════════════════════╗'));
     console.log(chalk.cyan('  ║       Welcome to ax-cli Setup Wizard           ║'));
     console.log(chalk.cyan('  ║         LOCAL/OFFLINE AI Assistant             ║'));
     console.log(chalk.cyan('  ╚════════════════════════════════════════════════╝\n'));
     console.log('  ax-cli is designed for LOCAL/OFFLINE AI inference.');
     console.log('  Run AI models locally without sending data to the cloud.\n');
-    console.log(chalk.dim('  For cloud providers, use dedicated CLIs:'));
-    console.log(chalk.dim('  • GLM (Z.AI):     npm install -g @defai.digital/ax-glm'));
-    console.log(chalk.dim('  • Grok (xAI):     npm install -g @defai.digital/ax-grok'));
-    console.log(chalk.dim('  • DeepSeek:       (coming soon) @defai.digital/ax-deepseek\n'));
-    // Handle --force flag: delete existing config
-    if (options.force && existsSync(AX_CLI_CONFIG_FILE)) {
-        console.log(chalk.yellow('  ⚠ Force flag detected - deleting existing configuration...'));
-        const deleted = deleteConfig();
-        if (deleted) {
-            console.log(chalk.green('  ✓ Existing configuration deleted\n'));
-        }
-        else {
-            console.log(chalk.red('  ✗ Failed to delete existing configuration\n'));
-            process.exit(1);
-        }
+    printCloudProviderInfo();
+}
+/**
+ * Print tier rankings (derived from MODEL_TIERS)
+ */
+function printTierRankings() {
+    console.log(chalk.bold('  2025 Offline Coding LLM Rankings (Updated Dec 2025):'));
+    for (const [tier, config] of Object.entries(MODEL_TIERS)) {
+        const newBadge = 'isNew' in config && config.isNew ? ' ' + chalk.yellow('★NEW') : '';
+        console.log(config.color(`    ${tier} ${config.displayName}`) +
+            chalk.dim(` (${config.rating})`) +
+            ` - ${config.label}: ${config.description}${newBadge}`);
     }
-    // Load existing config
-    const existingConfig = loadConfig();
     console.log();
-    printBoxHeader('Local/Offline Setup (Ollama, LMStudio, vLLM)', 53);
-    console.log('  Run AI models locally without an API key.\n');
-    console.log(chalk.bold('  2025 Offline Coding LLM Rankings (Updated Dec 2025):'));
-    console.log(chalk.green('    T1 Qwen 3') + chalk.dim(' (9.6/10)') + ' - PRIMARY: Best overall, coding leader');
-    console.log(chalk.magenta('    T2 GLM-4.6') + chalk.dim(' (9.4/10)') + ' - REFACTOR: Large-scale refactor + docs ' + chalk.yellow('★NEW'));
-    console.log(chalk.blue('    T3 DeepSeek') + chalk.dim(' (9.3/10)') + ' - SPEED: Quick patches, linting');
-    console.log(chalk.cyan('    T4 Codestral') + chalk.dim(' (8.4/10)') + ' - C++/RUST: Systems programming');
-    console.log(chalk.gray('    T5 Llama') + chalk.dim(' (8.1/10)') + ' - FALLBACK: Best compatibility\n');
-    // ═══════════════════════════════════════════════════════════════════
-    // STEP 1: Local Server Detection
-    // ═══════════════════════════════════════════════════════════════════
+}
+/**
+ * Handle force flag - delete existing config
+ */
+function handleForceFlag(force) {
+    if (!force || !existsSync(AX_CLI_CONFIG_FILE))
+        return;
+    console.log(chalk.yellow('  ⚠ Force flag detected - deleting existing configuration...'));
+    const deleted = deleteConfig();
+    if (deleted) {
+        console.log(chalk.green('  ✓ Existing configuration deleted\n'));
+    }
+    else {
+        console.log(chalk.red('  ✗ Failed to delete existing configuration\n'));
+        process.exit(1);
+    }
+}
+/**
+ * Step 1: Select local server
+ */
+async function selectLocalServer(existingConfig) {
     console.log(chalk.bold.cyan('\n  Step 1/3 — Local Server\n'));
-    const detectSpinner = ora('Detecting local inference servers...').start();
-    const detectedServers = await detectLocalServers();
+    const detectedServers = await withSpinner('Detecting local inference servers...', detectLocalServers);
     const availableServers = detectedServers.filter(s => s.available);
-    detectSpinner.stop();
-    let selectedBaseURL;
+    const defaultServerURL = normalizeBaseURL(existingConfig.baseURL || DEFAULT_LOCAL_BASE_URL);
     if (availableServers.length > 0) {
         console.log(chalk.green(`  ✓ Found ${availableServers.length} running server(s)\n`));
         const serverChoices = [
             ...availableServers.map(s => ({
                 name: `${chalk.green('●')} ${s.name} - ${s.url}`,
-                value: s.url,
+                value: normalizeBaseURL(s.url),
             })),
-            {
-                name: `${chalk.dim('○')} Enter custom URL...`,
-                value: '__custom__',
-            },
+            { name: `${chalk.dim('○')} Enter custom URL...`, value: '__custom__' },
         ];
         const serverSelection = await select({
             message: 'Select your local server:',
             choices: serverChoices,
-            default: existingConfig.baseURL || availableServers[0]?.url,
+            default: normalizeBaseURL(existingConfig.baseURL || availableServers[0]?.url),
         });
         if (serverSelection === '__custom__') {
-            selectedBaseURL = await input({
+            return normalizeBaseURL(await input({
                 message: 'Enter server URL:',
-                default: 'http://localhost:11434/v1',
+                default: defaultServerURL,
                 validate: validateURL,
-            });
-        }
-        else {
-            selectedBaseURL = serverSelection;
+            }));
         }
+        return normalizeBaseURL(serverSelection);
     }
-    else {
-        console.log(chalk.yellow('  No running servers detected.\n'));
-        console.log('  Common local server URLs:');
-        LOCAL_SERVERS.forEach(s => {
-            console.log(`    ${chalk.dim('•')} ${s.name}: ${chalk.dim(s.url)}`);
-        });
-        console.log();
-        console.log(chalk.dim('  Tip: Start Ollama with: ollama serve'));
-        console.log();
-        selectedBaseURL = await input({
-            message: 'Enter your server URL:',
-            default: existingConfig.baseURL || 'http://localhost:11434/v1',
-            validate: validateURL,
-        });
-    }
-    // ═══════════════════════════════════════════════════════════════════
-    // STEP 2: Model Selection (fetch from server or use defaults)
-    // ═══════════════════════════════════════════════════════════════════
+    // No servers detected
+    console.log(chalk.yellow('  No running servers detected.\n'));
+    console.log('  Common local server URLs:');
+    LOCAL_SERVERS.forEach(s => {
+        console.log(`    ${chalk.dim('•')} ${s.name}: ${chalk.dim(s.url)}`);
+    });
+    console.log();
+    console.log(chalk.dim('  Tip: Start Ollama with: ollama serve'));
+    console.log();
+    return normalizeBaseURL(await input({
+        message: 'Enter your server URL:',
+        default: defaultServerURL,
+        validate: validateURL,
+    }));
+}
+/**
+ * Step 2: Select model
+ */
+async function selectModel(baseURL, existingConfig) {
     console.log(chalk.bold.cyan('\n  Step 2/3 — Choose Model\n'));
-    const modelSpinner = ora('Fetching available models...').start();
-    let availableModels = await fetchLocalModels(selectedBaseURL);
-    modelSpinner.stop();
-    let selectedModel;
+    let availableModels = await withSpinner('Fetching available models...', () => fetchLocalModels(baseURL));
     if (availableModels.length > 0) {
         console.log(chalk.green(`  ✓ Found ${availableModels.length} model(s) on server\n`));
         const sortedModels = sortModelsByTier(availableModels);
         const modelChoices = buildModelChoices(sortedModels, true);
-        selectedModel = await selectModelWithCustom(modelChoices, existingConfig.defaultModel || sortedModels[0]?.id, PROVIDER_INFO.defaultModel);
+        const selectedModel = await selectModelWithCustom(modelChoices, existingConfig.defaultModel || sortedModels[0]?.id, PROVIDER_INFO.defaultModel);
+        return { model: selectedModel, models: availableModels };
     }
-    else {
-        console.log(chalk.yellow('  Could not fetch models from server.\n'));
-        console.log('  Recommended models by tier (Updated Dec 2025):\n');
-        console.log(chalk.bold.green('  T1 Qwen 3 (9.6/10) - PRIMARY:'));
-        LOCAL_QWEN_MODELS.slice(0, 3).forEach(m => {
-            console.log(`    ${chalk.green('•')} ${m.id}: ${chalk.dim(m.description)}`);
-        });
-        console.log(chalk.bold.magenta('\n  T2 GLM-4.6 (9.4/10) - REFACTOR + DOCS ★NEW:'));
-        LOCAL_GLM_MODELS.slice(0, 3).forEach(m => {
-            console.log(`    ${chalk.magenta('•')} ${m.id}: ${chalk.dim(m.description)}`);
-        });
-        console.log(chalk.bold.blue('\n  T3 DeepSeek (9.3/10) - SPEED:'));
-        LOCAL_DEEPSEEK_MODELS.slice(0, 2).forEach(m => {
-            console.log(`    ${chalk.blue('•')} ${m.id}: ${chalk.dim(m.description)}`);
-        });
-        console.log(chalk.bold.gray('\n  T5 Llama (8.1/10) - FALLBACK:'));
-        LOCAL_LLAMA_MODELS.slice(0, 2).forEach(m => {
-            console.log(`    ${chalk.gray('•')} ${m.id}: ${chalk.dim(m.description)}`);
-        });
-        console.log();
-        // ALL_LOCAL_MODELS already has tier prefixes from mapping, so don't add again
-        const modelChoices = buildModelChoices(ALL_LOCAL_MODELS, false);
-        selectedModel = await selectModelWithCustom(modelChoices, existingConfig.defaultModel || PROVIDER_INFO.defaultModel, PROVIDER_INFO.defaultModel);
-        availableModels = PROVIDER_INFO.models; // Fallback to predefined local models
-    }
-    // ═══════════════════════════════════════════════════════════════════
-    // STEP 3: Quick Setup Option
-    // ═══════════════════════════════════════════════════════════════════
+    // No models from server - use defaults
+    console.log(chalk.yellow('  Could not fetch models from server.\n'));
+    printModelRecommendations();
+    const modelChoices = buildModelChoices(ALL_LOCAL_MODELS, false);
+    const selectedModel = await selectModelWithCustom(modelChoices, existingConfig.defaultModel || PROVIDER_INFO.defaultModel, PROVIDER_INFO.defaultModel);
+    return { model: selectedModel, models: PROVIDER_INFO.models };
+}
+/**
+ * Step 3: Quick setup confirmation
+ */
+async function confirmQuickSetup(baseURL) {
     console.log(chalk.bold.cyan('\n  Step 3/3 — Quick Setup\n'));
     const useDefaults = await confirm({
         message: 'Use default settings for everything else? (Recommended)',
@@ -502,50 +453,56 @@ export async function runSetup(options = {}) {
     });
     if (useDefaults) {
         console.log(chalk.green('\n  ✓ Using default settings\n'));
+        return true;
+    }
+    // Validate server connection
+    const validateSpinner = ora('Validating local server connection...').start();
+    const isValid = await checkLocalServer(baseURL);
+    if (isValid) {
+        validateSpinner.succeed('Local server connection validated!');
     }
     else {
-        // Only validate if user wants detailed setup
-        const validateSpinner = ora('Validating local server connection...').start();
-        const isValid = await checkLocalServer(selectedBaseURL);
-        if (isValid) {
-            validateSpinner.succeed('Local server connection validated!');
-        }
-        else {
-            validateSpinner.warn('Server not responding (will save anyway)');
-            console.log(chalk.dim('\n  Tip: Make sure your local server is running before using ax-cli'));
-        }
+        validateSpinner.warn('Server not responding (will save anyway)');
+        console.log(chalk.dim('\n  Tip: Make sure your local server is running before using ax-cli'));
     }
-    // Save configuration
-    const newConfig = {
-        ...existingConfig,
-        selectedProvider: 'local',
-        serverType: 'local',
-        apiKey: '', // No API key needed for local
-        baseURL: selectedBaseURL,
-        defaultModel: selectedModel,
-        currentModel: selectedModel,
-        maxTokens: existingConfig.maxTokens ?? 8192,
-        temperature: existingConfig.temperature ?? 0.7,
-        // Ensure the selected model is always persisted, even if it wasn't in the fetched list
-        models: Array.from(new Set([...availableModels.map(m => m.id), selectedModel])),
-        _provider: PROVIDER_INFO.name,
-        _website: PROVIDER_INFO.website,
-        _isLocalServer: true,
-    };
-    saveConfig(newConfig);
-    console.log(chalk.green('\n  ✓ Configuration saved!\n'));
-    // ═══════════════════════════════════════════════════════════════════
-    // AutomatosX Integration (quick setup installs and configures by default)
-    // ═══════════════════════════════════════════════════════════════════
-    if (useDefaults) {
-        // Quick setup - install AutomatosX and run ax setup -f by default
+    return false;
+}
+/**
+ * Print setup summary
+ */
+function printSetupSummary(config) {
+    printBoxHeader('Configuration Summary', 41);
+    console.log(`  Provider:    ${config._provider}`);
+    console.log(`  Server:      ${config.baseURL}`);
+    console.log(`  Model:       ${config.defaultModel}`);
+    console.log(`  Config:      ${AX_CLI_CONFIG_FILE}`);
+    console.log();
+}
+/**
+ * Print next steps
+ */
+function printNextSteps() {
+    printBoxHeader('Next Steps', 41);
+    console.log(`  1. Run ${chalk.bold('ax-cli')} to start`);
+    console.log(`  2. Run ${chalk.bold('ax-cli --help')} for all options`);
+    console.log();
+    console.log(chalk.dim('  Note: Make sure your local server is running before using ax-cli'));
+    console.log();
+    printCloudProviderInfo();
+    console.log(chalk.green('  ✓ Setup complete! Happy coding!\n'));
+}
+/**
+ * Handle AutomatosX integration
+ */
+function handleAutomatosXIntegration(useQuickSetup) {
+    if (useQuickSetup) {
         let axStatus = getAutomatosXStatus();
         if (!axStatus.installed) {
             const installSpinner = ora('Installing AutomatosX for multi-agent AI orchestration...').start();
-            const installed = await installAutomatosX();
+            const installed = installAutomatosX();
             if (installed) {
                 installSpinner.succeed('AutomatosX installed successfully!');
-                axStatus = getAutomatosXStatus(); // Refresh status after install
+                axStatus = getAutomatosXStatus();
             }
             else {
                 installSpinner.stop();
@@ -556,10 +513,9 @@ export async function runSetup(options = {}) {
         else {
             console.log(chalk.green(`  ✓ AutomatosX detected${axStatus.version ? ` (v${axStatus.version})` : ''}`));
         }
-        // Run ax setup -f to configure AutomatosX with defaults
         if (axStatus.installed) {
             const setupSpinner = ora('Configuring AutomatosX...').start();
-            const setupSuccess = await runAutomatosXSetup();
+            const setupSuccess = runAutomatosXSetup();
             if (setupSuccess) {
                 setupSpinner.succeed('AutomatosX configured successfully!');
             }
@@ -569,37 +525,83 @@ export async function runSetup(options = {}) {
                 console.log(chalk.dim('  Configure manually later: ax setup -f'));
             }
         }
+        return;
+    }
+    // Non-quick setup - just show info
+    const axStatus = getAutomatosXStatus();
+    if (axStatus.installed) {
+        console.log(chalk.green(`  ✓ AutomatosX detected${axStatus.version ? ` (v${axStatus.version})` : ''}`));
     }
     else {
-        // Detailed setup - just show info
-        const axStatus = getAutomatosXStatus();
-        if (axStatus.installed) {
-            console.log(chalk.green(`  ✓ AutomatosX detected${axStatus.version ? ` (v${axStatus.version})` : ''}`));
-        }
-        else {
-            console.log(chalk.dim('  AutomatosX not installed. Install later: npm install -g @defai.digital/automatosx'));
-        }
+        console.log(chalk.dim('  AutomatosX not installed. Install later: npm install -g @defai.digital/automatosx'));
     }
-    // Show summary
-    printBoxHeader('Configuration Summary', 41);
-    console.log(`  Provider:    ${newConfig._provider}`);
-    console.log(`  Server:      ${newConfig.baseURL}`);
-    console.log(`  Model:       ${newConfig.defaultModel}`);
-    console.log(`  Config:      ${AX_CLI_CONFIG_FILE}`);
-    console.log();
-    // Show next steps
-    console.log();
-    printBoxHeader('Next Steps', 41);
-    console.log(`  1. Run ${chalk.bold('ax-cli')} to start`);
-    console.log(`  2. Run ${chalk.bold('ax-cli --help')} for all options`);
-    console.log();
-    console.log(chalk.dim('  Note: Make sure your local server is running before using ax-cli'));
+}
+/**
+ * Build and save configuration
+ */
+function buildAndSaveConfig(existingConfig, baseURL, model, models) {
+    const normalizedBaseURL = normalizeBaseURL(baseURL);
+    const newConfig = {
+        ...existingConfig,
+        selectedProvider: 'local',
+        serverType: 'local',
+        apiKey: '',
+        baseURL: normalizedBaseURL,
+        defaultModel: model,
+        currentModel: model,
+        maxTokens: existingConfig.maxTokens ?? 8192,
+        temperature: existingConfig.temperature ?? 0.7,
+        models: Array.from(new Set([...models.map(m => m.id), model])),
+        _provider: PROVIDER_INFO.name,
+        _website: PROVIDER_INFO.website,
+        _isLocalServer: true,
+    };
+    saveConfig(newConfig);
+    console.log(chalk.green('\n  ✓ Configuration saved!\n'));
+    return newConfig;
+}
+/**
+ * Run the setup wizard
+ *
+ * Refactored to use extracted helper functions for clarity:
+ * - printWelcomeBanner() - Welcome message and cloud provider info
+ * - handleForceFlag() - Handle --force config reset
+ * - printTierRankings() - Display LLM tier rankings
+ * - selectLocalServer() - Step 1: Server selection
+ * - selectModel() - Step 2: Model selection
+ * - confirmQuickSetup() - Step 3: Quick/detailed setup
+ * - buildAndSaveConfig() - Save configuration
+ * - handleAutomatosXIntegration() - AutomatosX setup
+ * - printSetupSummary() - Configuration summary
+ * - printNextSteps() - Next steps and completion
+ */
+export async function runSetup(options = {}) {
+    // Welcome banner
+    printWelcomeBanner();
+    // Handle --force flag
+    handleForceFlag(options.force ?? false);
+    // Load existing config
+    const existingConfig = loadConfig();
+    // Show setup header with tier rankings
     console.log();
-    console.log(chalk.dim('  For cloud providers:'));
-    console.log(chalk.dim('  • GLM (Z.AI):     ax-glm setup'));
-    console.log(chalk.dim('  • Grok (xAI):     ax-grok setup'));
+    printBoxHeader('Local/Offline Setup (Ollama, LMStudio, vLLM)', 53);
+    console.log('  Run AI models locally without an API key.\n');
+    printTierRankings();
+    // Step 1: Select local server
+    const selectedBaseURL = await selectLocalServer(existingConfig);
+    const normalizedBaseURL = normalizeBaseURL(selectedBaseURL);
+    // Step 2: Select model
+    const { model: selectedModel, models: availableModels } = await selectModel(normalizedBaseURL, existingConfig);
+    // Step 3: Quick setup confirmation
+    const useQuickSetup = await confirmQuickSetup(normalizedBaseURL);
+    // Save configuration
+    const newConfig = buildAndSaveConfig(existingConfig, normalizedBaseURL, selectedModel, availableModels);
+    // AutomatosX integration
+    handleAutomatosXIntegration(useQuickSetup);
+    // Show summary and next steps
+    printSetupSummary(newConfig);
     console.log();
-    console.log(chalk.green('  ✓ Setup complete! Happy coding!\n'));
+    printNextSteps();
 }
 /**
  * Get the selected provider from config