@appkit/llamacpp-cli 1.11.0 → 1.12.1

package/src/commands/config.ts

@@ -1,323 +0,0 @@
-import chalk from 'chalk';
-import * as path from 'path';
-import * as fs from 'fs/promises';
-import { stateManager } from '../lib/state-manager';
-import { statusChecker } from '../lib/status-checker';
-import { launchctlManager } from '../lib/launchctl-manager';
-import { configGenerator } from '../lib/config-generator';
-import { autoRotateIfNeeded } from '../utils/log-utils';
-import { modelScanner } from '../lib/model-scanner';
-import { sanitizeModelName } from '../types/server-config';
-import { getLogsDir, getLaunchAgentsDir } from '../utils/file-utils';
-
-export interface ConfigUpdateOptions {
-  model?: string;
-  host?: string;
-  threads?: number;
-  ctxSize?: number;
-  gpuLayers?: number;
-  verbose?: boolean;
-  flags?: string;
-  restart?: boolean;
-}
-
-export async function serverConfigCommand(
-  identifier: string,
-  options: ConfigUpdateOptions
-): Promise<void> {
-  // Find the server
-  const server = await stateManager.findServer(identifier);
-
-  if (!server) {
-    console.error(chalk.red(`❌ Server not found: ${identifier}`));
-    console.log(chalk.dim('\nAvailable servers:'));
-    const allServers = await stateManager.getAllServers();
-    if (allServers.length === 0) {
-      console.log(chalk.dim('  (none)'));
-      console.log(chalk.dim('\nCreate a server: llamacpp server create <model-filename>'));
-    } else {
-      allServers.forEach(s => {
-        console.log(chalk.dim(`  - ${s.id} (port ${s.port})`));
-      });
-    }
-    process.exit(1);
-  }
-
-  // Check if any config options were provided
-  const hasChanges = options.model !== undefined ||
-                     options.host !== undefined ||
-                     options.threads !== undefined ||
-                     options.ctxSize !== undefined ||
-                     options.gpuLayers !== undefined ||
-                     options.verbose !== undefined ||
-                     options.flags !== undefined;
-
-  if (!hasChanges) {
-    console.error(chalk.red('❌ No configuration changes specified'));
-    console.log(chalk.dim('\nAvailable options:'));
-    console.log(chalk.dim('  --model <filename>  Model filename or path'));
-    console.log(chalk.dim('  --host <address>    Bind address (127.0.0.1 or 0.0.0.0)'));
-    console.log(chalk.dim('  --threads <n>       Number of threads'));
-    console.log(chalk.dim('  --ctx-size <n>      Context size'));
-    console.log(chalk.dim('  --gpu-layers <n>    GPU layers'));
-    console.log(chalk.dim('  --verbose           Enable verbose logging'));
-    console.log(chalk.dim('  --no-verbose        Disable verbose logging'));
-    console.log(chalk.dim('  --flags <flags>     Custom llama-server flags (comma-separated)'));
-    console.log(chalk.dim('  --restart           Auto-restart if running'));
-    console.log(chalk.dim('\nExamples:'));
-    console.log(chalk.dim(`  llamacpp server config ${server.id} --model llama-3.2-1b.gguf --restart`));
-    console.log(chalk.dim(`  llamacpp server config ${server.id} --ctx-size 8192 --restart`));
-    console.log(chalk.dim(`  llamacpp server config ${server.id} --flags="--pooling,mean" --restart`));
-    process.exit(1);
-  }
-
-  // Resolve model path if model option is provided
-  let newModelPath: string | undefined;
-  let newModelName: string | undefined;
-  let newServerId: string | undefined;
-  let isModelMigration = false;
-
-  if (options.model !== undefined) {
-    const resolvedPath = await modelScanner.resolveModelPath(options.model);
-    if (!resolvedPath) {
-      console.error(chalk.red(`❌ Model not found: ${options.model}`));
-      console.log(chalk.dim('\nRun: llamacpp ls'));
-      process.exit(1);
-    }
-    newModelPath = resolvedPath;
-    newModelName = path.basename(resolvedPath);
-    newServerId = sanitizeModelName(newModelName);
-
-    // Check if this is a model migration (ID will change)
-    if (newServerId !== server.id) {
-      isModelMigration = true;
-
-      // Check for ID conflict
-      const existingServer = await stateManager.loadServerConfig(newServerId);
-      if (existingServer) {
-        console.error(chalk.red(`❌ A server with ID "${newServerId}" already exists`));
-        console.log(chalk.dim('\nChanging the model would create this server ID, but it conflicts with an existing server.'));
-        console.log(chalk.dim('Delete the existing server first:'));
-        console.log(chalk.dim(`  llamacpp server rm ${newServerId}`));
-        process.exit(1);
-      }
-
-      console.log(chalk.yellow('⚠️  Changing the model will migrate to a new server ID'));
-      console.log(chalk.dim(`   Old ID: ${server.id}`));
-      console.log(chalk.dim(`   New ID: ${newServerId}\n`));
-    }
-  }
-
-  // Check current status
-  const updatedServer = await statusChecker.updateServerStatus(server);
-  const wasRunning = updatedServer.status === 'running';
-
-  if (wasRunning && !options.restart) {
-    console.warn(chalk.yellow('⚠️  Server is currently running'));
-    console.log(chalk.dim('Changes will require a restart to take effect.'));
-    console.log(chalk.dim('Use --restart flag to automatically restart the server.\n'));
-  }
-
-  // Show what will change
-  console.log(chalk.bold('Configuration Changes:'));
-  console.log('─'.repeat(70));
-
-  if (newModelPath !== undefined && newModelName !== undefined) {
-    const oldModelName = path.basename(server.modelPath);
-    console.log(`${chalk.bold('Model:')}          ${chalk.dim(oldModelName)} → ${chalk.green(newModelName)}`);
-    console.log(`${chalk.dim('                  ')}${chalk.dim(server.modelPath)}`);
-    console.log(`${chalk.dim('                  ')}${chalk.dim(newModelPath)}`);
-    if (isModelMigration && newServerId) {
-      console.log(`${chalk.bold('Server ID:')}      ${chalk.dim(server.id)} → ${chalk.green(newServerId)}`);
-    }
-  }
-  if (options.host !== undefined) {
-    console.log(`${chalk.bold('Host:')}           ${chalk.dim(server.host)} → ${chalk.green(options.host)}`);
-
-    // Security warning for 0.0.0.0
-    if (options.host === '0.0.0.0') {
-      console.log(chalk.yellow('\n⚠️  WARNING: Binding to 0.0.0.0 allows remote access from any network interface.'));
-      console.log(chalk.yellow('   This exposes your server to your local network and potentially the internet.'));
-      console.log(chalk.yellow('   Use 127.0.0.1 for localhost-only access (recommended for local development).\n'));
-    }
-  }
-  if (options.threads !== undefined) {
-    console.log(`${chalk.bold('Threads:')}        ${chalk.dim(server.threads.toString())} → ${chalk.green(options.threads.toString())}`);
-  }
-  if (options.ctxSize !== undefined) {
-    console.log(`${chalk.bold('Context Size:')}   ${chalk.dim(server.ctxSize.toLocaleString())} → ${chalk.green(options.ctxSize.toLocaleString())}`);
-  }
-  if (options.gpuLayers !== undefined) {
-    console.log(`${chalk.bold('GPU Layers:')}     ${chalk.dim(server.gpuLayers.toString())} → ${chalk.green(options.gpuLayers.toString())}`);
-  }
-  if (options.verbose !== undefined) {
-    const oldValue = server.verbose ? 'enabled' : 'disabled';
-    const newValue = options.verbose ? 'enabled' : 'disabled';
-    console.log(`${chalk.bold('Verbose Logs:')}   ${chalk.dim(oldValue)} → ${chalk.green(newValue)}`);
-  }
-  if (options.flags !== undefined) {
-    const oldValue = server.customFlags?.join(' ') || 'none';
-    const newValue = options.flags || 'none';
-    console.log(`${chalk.bold('Custom Flags:')}   ${chalk.dim(oldValue)} → ${chalk.green(newValue)}`);
-  }
-  console.log('');
-
-  // Parse custom flags if provided
-  let customFlags: string[] | undefined;
-  if (options.flags !== undefined) {
-    if (options.flags === '') {
-      customFlags = undefined;
-    } else {
-      customFlags = options.flags.split(',').map(f => f.trim()).filter(f => f.length > 0);
-    }
-  }
-
-  // Handle model migration (different code path when server ID changes)
-  if (isModelMigration && newServerId && newModelPath && newModelName) {
-    // Stop old server if running
-    if (wasRunning) {
-      console.log(chalk.dim('Stopping old server...'));
-      await launchctlManager.unloadService(server.plistPath);
-      await new Promise(resolve => setTimeout(resolve, 1000));
-    }
-
-    // Delete old plist and config
-    console.log(chalk.dim('Removing old server configuration...'));
-    try {
-      await fs.unlink(server.plistPath);
-    } catch (err) {
-      // Plist might not exist, that's ok
-    }
-    await stateManager.deleteServerConfig(server.id);
-
-    // Create new config with new ID and all settings
-    const logsDir = getLogsDir();
-    const plistDir = getLaunchAgentsDir();
-
-    const newConfig = {
-      ...server,
-      id: newServerId,
-      modelPath: newModelPath,
-      modelName: newModelName,
-      ...(options.host !== undefined && { host: options.host }),
-      ...(options.threads !== undefined && { threads: options.threads }),
-      ...(options.ctxSize !== undefined && { ctxSize: options.ctxSize }),
-      ...(options.gpuLayers !== undefined && { gpuLayers: options.gpuLayers }),
-      ...(options.verbose !== undefined && { verbose: options.verbose }),
-      ...(options.flags !== undefined && { customFlags }),
-      // Update plist-related paths
-      label: `com.llama.${newServerId}`,
-      plistPath: path.join(plistDir, `com.llama.${newServerId}.plist`),
-      stdoutPath: path.join(logsDir, `${newServerId}.stdout`),
-      stderrPath: path.join(logsDir, `${newServerId}.stderr`),
-      status: 'stopped' as const,
-      pid: undefined,
-      lastStopped: new Date().toISOString(),
-    };
-
-    console.log(chalk.dim('Creating new server configuration...'));
-    await stateManager.saveServerConfig(newConfig);
-    await launchctlManager.createPlist(newConfig);
-
-    // Start new server if restart flag is set
-    if (wasRunning && options.restart) {
-      console.log(chalk.dim('Starting new server...'));
-      await launchctlManager.loadService(newConfig.plistPath);
-      await launchctlManager.startService(newConfig.label);
-
-      // Wait and verify
-      await new Promise(resolve => setTimeout(resolve, 2000));
-      const finalStatus = await statusChecker.updateServerStatus(newConfig);
-
-      if (finalStatus.status === 'running') {
-        console.log(chalk.green(`✅ Server migrated successfully to new ID: ${newServerId}`));
-        console.log(chalk.dim(`   Port: http://localhost:${finalStatus.port}`));
-        if (finalStatus.pid) {
-          console.log(chalk.dim(`   PID: ${finalStatus.pid}`));
-        }
-      } else {
-        console.error(chalk.red('❌ Server failed to start with new configuration'));
-        console.log(chalk.dim('   Check logs: ') + `llamacpp server logs ${newServerId} --errors`);
-        process.exit(1);
-      }
-    } else {
-      console.log(chalk.green(`✅ Server migrated successfully to new ID: ${newServerId}`));
-      if (!wasRunning) {
-        console.log(chalk.dim('\n   Start server: ') + `llamacpp server start ${newServerId}`);
-      } else {
-        console.log(chalk.dim('\n   Start server: ') + `llamacpp server start ${newServerId}`);
-      }
-    }
-
-    return; // Exit early for migration path
-  }
-
-  // Normal config update (no model migration)
-  // Unload service if running and restart flag is set
-  if (wasRunning && options.restart) {
-    console.log(chalk.dim('Stopping server...'));
-    await launchctlManager.unloadService(server.plistPath);
-    await new Promise(resolve => setTimeout(resolve, 1000));
-  }
-
-  const updatedConfig = {
-    ...server,
-    ...(newModelPath !== undefined && { modelPath: newModelPath, modelName: newModelName }),
-    ...(options.host !== undefined && { host: options.host }),
-    ...(options.threads !== undefined && { threads: options.threads }),
-    ...(options.ctxSize !== undefined && { ctxSize: options.ctxSize }),
-    ...(options.gpuLayers !== undefined && { gpuLayers: options.gpuLayers }),
-    ...(options.verbose !== undefined && { verbose: options.verbose }),
-    ...(options.flags !== undefined && { customFlags }),
-  };
-
-  await stateManager.updateServerConfig(server.id, updatedConfig);
-
-  console.log(chalk.dim('Regenerating service configuration...'));
-  await launchctlManager.createPlist(updatedConfig);
-
-  // Restart server if it was running and restart flag is set
-  if (wasRunning && options.restart) {
-    // Auto-rotate logs if they exceed 100MB
-    try {
-      const result = await autoRotateIfNeeded(updatedConfig.stdoutPath, updatedConfig.stderrPath, 100);
-      if (result.rotated) {
-        console.log(chalk.dim('Auto-rotated large log files:'));
-        for (const file of result.files) {
-          console.log(chalk.dim(`  → ${file}`));
-        }
-      }
-    } catch (error) {
-      // Non-fatal, just warn
-      console.log(chalk.yellow(`⚠️  Failed to rotate logs: ${(error as Error).message}`));
-    }
-
-    console.log(chalk.dim('Starting server with new configuration...'));
-    await launchctlManager.loadService(updatedConfig.plistPath);
-    await launchctlManager.startService(updatedConfig.label);
-
-    // Wait and verify
-    await new Promise(resolve => setTimeout(resolve, 2000));
-    const finalStatus = await statusChecker.updateServerStatus(updatedConfig);
-
-    if (finalStatus.status === 'running') {
-      console.log(chalk.green(`✅ Server restarted successfully with new configuration`));
-      console.log(chalk.dim(`   Port: http://localhost:${finalStatus.port}`));
-      if (finalStatus.pid) {
-        console.log(chalk.dim(`   PID: ${finalStatus.pid}`));
-      }
-    } else {
-      console.error(chalk.red('❌ Server failed to start with new configuration'));
-      console.log(chalk.dim('   Check logs: ') + `llamacpp server logs ${server.id} --errors`);
-      process.exit(1);
-    }
-  } else {
-    console.log(chalk.green('✅ Configuration updated successfully'));
-    if (wasRunning && !options.restart) {
-      console.log(chalk.yellow('\n⚠️  Server is still running with old configuration'));
-      console.log(chalk.dim('   Restart to apply changes: ') + `llamacpp server stop ${server.id} && llamacpp server start ${server.id}`);
-    } else if (!wasRunning) {
-      console.log(chalk.dim('\n   Start server: ') + `llamacpp server start ${server.id}`);
-    }
-  }
-}

package/src/commands/create.ts

@@ -1,183 +0,0 @@
-import chalk from 'chalk';
-import * as path from 'path';
-import * as fs from 'fs';
-import { modelScanner } from '../lib/model-scanner';
-import { stateManager } from '../lib/state-manager';
-import { configGenerator, ServerOptions } from '../lib/config-generator';
-import { portManager } from '../lib/port-manager';
-import { launchctlManager } from '../lib/launchctl-manager';
-import { statusChecker } from '../lib/status-checker';
-import { commandExists } from '../utils/process-utils';
-import { formatBytes } from '../utils/format-utils';
-import { ensureDir, parseMetalMemoryFromLog } from '../utils/file-utils';
-import { ensureModelsDirectory } from '../lib/models-dir-setup';
-
-interface CreateOptions {
-  port?: number;
-  host?: string;
-  threads?: number;
-  ctxSize?: number;
-  gpuLayers?: number;
-  verbose?: boolean;
-  flags?: string;
-}
-
-export async function createCommand(model: string, options: CreateOptions): Promise<void> {
-  // Initialize state manager
-  await stateManager.initialize();
-
-  // 1. Check if llama-server exists
-  if (!(await commandExists('llama-server'))) {
-    throw new Error('llama-server not found. Install with: brew install llama.cpp');
-  }
-
-  // 2. Ensure models directory exists if model is not an absolute path
-  if (!path.isAbsolute(model)) {
-    const modelsDir = await stateManager.getModelsDirectory();
-    if (!fs.existsSync(modelsDir)) {
-      await ensureModelsDirectory();
-    }
-  }
-
-  // 3. Resolve model path
-  const modelPath = await modelScanner.resolveModelPath(model);
-  if (!modelPath) {
-    throw new Error(`Model not found: ${model}\n\nRun: llamacpp ls`);
-  }
-
-  const modelName = path.basename(modelPath);
-
-  // 4. Check if server already exists for this model
-  const existingServer = await stateManager.serverExistsForModel(modelPath);
-  if (existingServer) {
-    throw new Error(`Server already exists for ${modelName}\n\nUse: llamacpp server start ${modelName}`);
-  }
-
-  // 5. Get model size
-  const modelSize = await modelScanner.getModelSize(modelName);
-  if (!modelSize) {
-    throw new Error(`Failed to read model file: ${modelPath}`);
-  }
-
-  // 6. Determine port
-  let port: number;
-  if (options.port) {
-    portManager.validatePort(options.port);
-    const available = await portManager.isPortAvailable(options.port);
-    if (!available) {
-      throw new Error(`Port ${options.port} is already in use`);
-    }
-    port = options.port;
-  } else {
-    port = await portManager.findAvailablePort();
-  }
-
-  // 7. Generate server configuration
-  console.log(chalk.blue(`🚀 Creating server for ${modelName}\n`));
-
-  // Parse custom flags if provided
-  let customFlags: string[] | undefined;
-  if (options.flags) {
-    customFlags = options.flags.split(',').map(f => f.trim()).filter(f => f.length > 0);
-  }
-
-  const serverOptions: ServerOptions = {
-    port: options.port,
-    host: options.host,
-    threads: options.threads,
-    ctxSize: options.ctxSize,
-    gpuLayers: options.gpuLayers,
-    verbose: options.verbose,
-    customFlags,
-  };
-
-  const config = await configGenerator.generateConfig(
-    modelPath,
-    modelName,
-    modelSize,
-    port,
-    serverOptions
-  );
-
-  // Security warning for 0.0.0.0
-  if (config.host === '0.0.0.0') {
-    console.log(chalk.yellow('⚠️  WARNING: Binding to 0.0.0.0 allows remote access from any network interface.'));
-    console.log(chalk.yellow('   This exposes your server to your local network and potentially the internet.'));
-    console.log(chalk.yellow('   Use 127.0.0.1 for localhost-only access (recommended for local development).\n'));
-  }
-
-  // Display configuration
-  console.log(chalk.dim(`Model: ${modelPath}`));
-  console.log(chalk.dim(`Size: ${formatBytes(modelSize)}`));
-  console.log(chalk.dim(`Host: ${config.host}`));
-  console.log(chalk.dim(`Port: ${config.port}${options.port ? '' : ' (auto-assigned)'}`));
-  console.log(chalk.dim(`Threads: ${config.threads}`));
-  console.log(chalk.dim(`Context Size: ${config.ctxSize}`));
-  console.log(chalk.dim(`GPU Layers: ${config.gpuLayers}`));
-  console.log(chalk.dim(`Verbose Logging: ${config.verbose ? 'enabled' : 'disabled'}`));
-  if (config.customFlags && config.customFlags.length > 0) {
-    console.log(chalk.dim(`Custom Flags: ${config.customFlags.join(' ')}`));
-  }
-  console.log();
-
-  // 7. Ensure log directory exists
-  await ensureDir(path.dirname(config.stdoutPath));
-
-  // 8. Create plist file
-  console.log(chalk.dim('Creating launchctl service...'));
-  await launchctlManager.createPlist(config);
-
-  // 9. Load service
-  try {
-    await launchctlManager.loadService(config.plistPath);
-  } catch (error) {
-    // Clean up plist if load fails
-    await launchctlManager.deletePlist(config.plistPath);
-    throw new Error(`Failed to load service: ${(error as Error).message}`);
-  }
-
-  // 10. Start service
-  try {
-    await launchctlManager.startService(config.label);
-  } catch (error) {
-    // Clean up if start fails
-    await launchctlManager.unloadService(config.plistPath);
-    await launchctlManager.deletePlist(config.plistPath);
-    throw new Error(`Failed to start service: ${(error as Error).message}`);
-  }
-
-  // 11. Wait for startup
-  console.log(chalk.dim('Waiting for server to start...'));
-  const started = await launchctlManager.waitForServiceStart(config.label, 5000);
-
-  if (!started) {
-    // Clean up if startup fails
-    await launchctlManager.unloadService(config.plistPath);
-    await launchctlManager.deletePlist(config.plistPath);
-    throw new Error('Server failed to start. Check logs with: llamacpp server logs --errors');
-  }
-
-  // 12. Update config with running status
-  let updatedConfig = await statusChecker.updateServerStatus(config);
-
-  // 13. Parse Metal (GPU) memory allocation from logs
-  // Wait a few seconds for model to start loading (large models take time)
-  console.log(chalk.dim('Detecting Metal (GPU) memory allocation...'));
-  await new Promise(resolve => setTimeout(resolve, 8000)); // 8 second delay
-  const metalMemoryMB = await parseMetalMemoryFromLog(updatedConfig.stderrPath);
-  if (metalMemoryMB) {
-    updatedConfig = { ...updatedConfig, metalMemoryMB };
-    console.log(chalk.dim(`Metal memory: ${metalMemoryMB.toFixed(0)} MB`));
-  }
-
-  // 14. Save server config
-  await stateManager.saveServerConfig(updatedConfig);
-
-  // 15. Display success message
-  console.log();
-  console.log(chalk.green('✅ Server created and started successfully!'));
-  console.log();
-  console.log(chalk.dim(`Connect: http://${config.host}:${config.port}`));
-  console.log(chalk.dim(`View logs: llamacpp server logs ${config.id}`));
-  console.log(chalk.dim(`Stop: llamacpp server stop ${config.id}`));
-}

package/src/commands/delete.ts

@@ -1,74 +0,0 @@
-import chalk from 'chalk';
-import * as readline from 'readline';
-import { stateManager } from '../lib/state-manager';
-import { launchctlManager } from '../lib/launchctl-manager';
-
-export async function deleteCommand(identifier: string): Promise<void> {
-  // Find server
-  const server = await stateManager.findServer(identifier);
-  if (!server) {
-    throw new Error(`Server not found: ${identifier}\n\nUse: llamacpp ps`);
-  }
-
-  // Confirm deletion
-  console.log(chalk.yellow(`⚠️  Delete server configuration for ${server.modelName}?`));
-  console.log(chalk.dim('   This will remove the launchd service but keep the model file.'));
-  console.log();
-
-  const confirmed = await confirmDeletion();
-  if (!confirmed) {
-    console.log(chalk.dim('Cancelled'));
-    return;
-  }
-
-  console.log();
-  console.log(chalk.blue(`🗑️  Deleting server ${server.modelName}...`));
-
-  // Unload service (stops and removes from launchd)
-  if (server.status === 'running') {
-    console.log(chalk.dim('Stopping and unloading service...'));
-  } else {
-    console.log(chalk.dim('Unloading service...'));
-  }
-  try {
-    await launchctlManager.unloadService(server.plistPath);
-    if (server.status === 'running') {
-      await launchctlManager.waitForServiceStop(server.label, 5000);
-    }
-  } catch (error) {
-    console.log(chalk.yellow('⚠️  Failed to unload service gracefully'));
-  }
-
-  // Delete plist
-  console.log(chalk.dim('Deleting plist file...'));
-  await launchctlManager.deletePlist(server.plistPath);
-
-  // Delete server config
-  console.log(chalk.dim('Deleting server configuration...'));
-  await stateManager.deleteServerConfig(server.id);
-
-  // Success
-  console.log();
-  console.log(chalk.green('✅ Server deleted'));
-  console.log(chalk.dim(`   Plist removed: ${server.plistPath}`));
-  console.log(chalk.dim(`   Config removed`));
-  console.log();
-  console.log(chalk.dim(`   Model file preserved at: ${server.modelPath}`));
-}
-
-/**
- * Prompt user for confirmation
- */
-function confirmDeletion(): Promise<boolean> {
-  return new Promise((resolve) => {
-    const rl = readline.createInterface({
-      input: process.stdin,
-      output: process.stdout,
-    });
-
-    rl.question(chalk.yellow("   Type 'yes' to confirm: "), (answer) => {
-      rl.close();
-      resolve(answer.toLowerCase() === 'yes');
-    });
-  });
-}

package/src/commands/list.ts

@@ -1,37 +0,0 @@
-import chalk from 'chalk';
-import Table from 'cli-table3';
-import { modelScanner } from '../lib/model-scanner';
-import { formatBytes, formatDateShort } from '../utils/format-utils';
-import { stateManager } from '../lib/state-manager';
-
-export async function listCommand(): Promise<void> {
-  const modelsDir = await stateManager.getModelsDirectory();
-  console.log(chalk.blue(`📦 Available models in ${modelsDir}\n`));
-
-  const models = await modelScanner.scanModels();
-
-  if (models.length === 0) {
-    console.log(chalk.yellow('No GGUF models found.'));
-    console.log(chalk.dim(`\nDownload models with: llamacpp pull <repo> --file <filename>`));
-    return;
-  }
-
-  const table = new Table({
-    head: ['MODEL', 'SIZE', 'MODIFIED'],
-    colWidths: [50, 12, 15],
-  });
-
-  for (const model of models) {
-    table.push([
-      model.filename,
-      model.sizeFormatted,
-      formatDateShort(model.modified),
-    ]);
-  }
-
-  console.log(table.toString());
-
-  const totalSize = models.reduce((sum, m) => sum + m.size, 0);
-  console.log(chalk.dim(`\nTotal: ${models.length} models (${formatBytes(totalSize)})`));
-  console.log(chalk.dim(`\nCreate a server: llamacpp server create <model-filename>`));
-}