@appkit/llamacpp-cli 1.11.0 → 1.12.1

package/src/commands/server-show.ts

@@ -1,161 +0,0 @@
-import chalk from 'chalk';
-import { stateManager } from '../lib/state-manager';
-import { statusChecker } from '../lib/status-checker';
-import { formatUptime, formatBytes, formatContextSize } from '../utils/format-utils';
-import { getProcessMemory } from '../utils/process-utils';
-import { getFileSize, formatFileSize, getArchivedLogInfo } from '../utils/log-utils';
-import { fileExists } from '../utils/file-utils';
-
-export async function serverShowCommand(identifier: string): 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 ${s.port})`));
-      });
-    }
-    process.exit(1);
-  }
-
-  // Update status to get real-time info
-  console.log(chalk.dim('Checking server status...\n'));
-  const updatedServer = await statusChecker.updateServerStatus(server);
-
-  // Display server information
-  console.log(chalk.bold('Server Configuration:'));
-  console.log('─'.repeat(70));
-
-  // Basic info
-  console.log(`${chalk.bold('Server ID:')}      ${updatedServer.id}`);
-  console.log(`${chalk.bold('Model Name:')}     ${updatedServer.modelName}`);
-  console.log(`${chalk.bold('Model Path:')}     ${chalk.dim(updatedServer.modelPath)}`);
-  console.log(`${chalk.bold('Host:')}           ${updatedServer.host}`);
-  console.log(`${chalk.bold('Port:')}           http://${updatedServer.host}:${updatedServer.port}`);
-
-  // Status with color
-  let statusText: string;
-  let statusColor: (text: string) => string;
-  switch (updatedServer.status) {
-    case 'running':
-      statusText = '✅ RUNNING';
-      statusColor = chalk.green;
-      break;
-    case 'crashed':
-      statusText = '❌ CRASHED';
-      statusColor = chalk.red;
-      break;
-    default:
-      statusText = '⚠️  STOPPED';
-      statusColor = chalk.yellow;
-  }
-  console.log(`${chalk.bold('Status:')}         ${statusColor(statusText)}`);
-
-  if (updatedServer.pid) {
-    console.log(`${chalk.bold('PID:')}            ${updatedServer.pid}`);
-  }
-
-  // Runtime info for running servers
-  if (updatedServer.status === 'running') {
-    if (updatedServer.lastStarted) {
-      const uptime = formatUptime(updatedServer.lastStarted);
-      console.log(`${chalk.bold('Uptime:')}         ${uptime}`);
-    }
-
-    if (updatedServer.pid) {
-      const cpuMemoryBytes = await getProcessMemory(updatedServer.pid);
-      if (cpuMemoryBytes !== null) {
-        const metalMemoryBytes = updatedServer.metalMemoryMB ? updatedServer.metalMemoryMB * 1024 * 1024 : 0;
-        const totalMemoryBytes = cpuMemoryBytes + metalMemoryBytes;
-
-        if (metalMemoryBytes > 0) {
-          console.log(`${chalk.bold('Memory:')}         ${formatBytes(totalMemoryBytes)} (CPU: ${formatBytes(cpuMemoryBytes)}, GPU: ${formatBytes(metalMemoryBytes)})`);
-        } else {
-          console.log(`${chalk.bold('Memory:')}         ${formatBytes(cpuMemoryBytes)} (CPU only)`);
-        }
-      }
-    }
-  }
-
-  // Configuration section
-  console.log('\n' + '─'.repeat(70));
-  console.log(chalk.bold('Configuration:'));
-  console.log('─'.repeat(70));
-  console.log(`${chalk.bold('Threads:')}        ${updatedServer.threads}`);
-  console.log(`${chalk.bold('Context Size:')}   ${formatContextSize(updatedServer.ctxSize)} (total)`);
-  console.log(`${chalk.bold('GPU Layers:')}     ${updatedServer.gpuLayers}`);
-  console.log(`${chalk.bold('Embeddings:')}     ${updatedServer.embeddings ? 'enabled' : 'disabled'}`);
-  console.log(`${chalk.bold('Jinja:')}          ${updatedServer.jinja ? 'enabled' : 'disabled'}`);
-  console.log(`${chalk.bold('Verbose Logs:')}   ${updatedServer.verbose ? chalk.green('enabled') : chalk.dim('disabled')}`);
-  if (updatedServer.customFlags && updatedServer.customFlags.length > 0) {
-    console.log(`${chalk.bold('Custom Flags:')}   ${updatedServer.customFlags.join(' ')}`);
-  }
-
-  // Logs section
-  console.log('\n' + '─'.repeat(70));
-  console.log(chalk.bold('Logs:'));
-  console.log('─'.repeat(70));
-
-  // Get current log sizes
-  const stderrSize = (await fileExists(updatedServer.stderrPath))
-    ? await getFileSize(updatedServer.stderrPath)
-    : 0;
-  const stdoutSize = (await fileExists(updatedServer.stdoutPath))
-    ? await getFileSize(updatedServer.stdoutPath)
-    : 0;
-
-  // Get archived log info
-  const archivedInfo = await getArchivedLogInfo(updatedServer.id);
-
-  console.log(`${chalk.bold('Stderr:')}         ${formatFileSize(stderrSize)} (current)`);
-  console.log(`${chalk.bold('Stdout:')}         ${formatFileSize(stdoutSize)} (current)`);
-
-  if (archivedInfo.count > 0) {
-    console.log(`${chalk.bold('Archived:')}       ${formatFileSize(archivedInfo.totalSize)} (${archivedInfo.count} file${archivedInfo.count > 1 ? 's' : ''})`);
-  }
-
-  // Timestamps section
-  console.log('\n' + '─'.repeat(70));
-  console.log(chalk.bold('Timestamps:'));
-  console.log('─'.repeat(70));
-  console.log(`${chalk.bold('Created:')}        ${new Date(updatedServer.createdAt).toLocaleString()}`);
-  if (updatedServer.lastStarted) {
-    console.log(`${chalk.bold('Last Started:')}   ${new Date(updatedServer.lastStarted).toLocaleString()}`);
-  }
-  if (updatedServer.lastStopped) {
-    console.log(`${chalk.bold('Last Stopped:')}   ${new Date(updatedServer.lastStopped).toLocaleString()}`);
-  }
-
-  // System paths section
-  console.log('\n' + '─'.repeat(70));
-  console.log(chalk.bold('System Paths:'));
-  console.log('─'.repeat(70));
-  console.log(`${chalk.bold('Service Label:')}  ${updatedServer.label}`);
-  console.log(`${chalk.bold('Plist File:')}     ${chalk.dim(updatedServer.plistPath)}`);
-  console.log(`${chalk.bold('Stdout Log:')}     ${chalk.dim(updatedServer.stdoutPath)}`);
-  console.log(`${chalk.bold('Stderr Log:')}     ${chalk.dim(updatedServer.stderrPath)}`);
-
-  // Helpful commands
-  console.log('\n' + '─'.repeat(70));
-  console.log(chalk.bold('Quick Commands:'));
-  console.log('─'.repeat(70));
-
-  if (updatedServer.status === 'running') {
-    console.log(chalk.dim('  View logs:        ') + `llamacpp server logs ${updatedServer.id}`);
-    console.log(chalk.dim('  Interactive chat: ') + `llamacpp server run ${updatedServer.id}`);
-    console.log(chalk.dim('  Stop server:      ') + `llamacpp server stop ${updatedServer.id}`);
-  } else {
-    console.log(chalk.dim('  Start server:     ') + `llamacpp server start ${updatedServer.id}`);
-    if (updatedServer.status === 'crashed') {
-      console.log(chalk.dim('  View error logs:  ') + `llamacpp server logs ${updatedServer.id} --errors`);
-    }
-  }
-  console.log(chalk.dim('  Remove server:    ') + `llamacpp server rm ${updatedServer.id}`);
-}

package/src/commands/show.ts

@@ -1,207 +0,0 @@
-import chalk from 'chalk';
-import Table from 'cli-table3';
-import { modelSearch } from '../lib/model-search';
-import { modelDownloader } from '../lib/model-downloader';
-import { formatBytes } from '../utils/format-utils';
-import * as https from 'https';
-
-interface ShowOptions {
-  file?: string;
-}
-
-interface ModelDetails {
-  modelId: string;
-  author: string;
-  modelName: string;
-  downloads: number;
-  likes: number;
-  lastModified: string;
-  tags: string[];
-  library?: string;
-  license?: string;
-}
-
-interface FileDetails {
-  filename: string;
-  size: number;
-  lfs?: {
-    oid: string;
-    size: number;
-  };
-}
-
-export async function showCommand(identifier: string, options: ShowOptions): Promise<void> {
-  // Parse identifier
-  const parsed = modelDownloader.parseHFIdentifier(identifier);
-  const filename = options.file || parsed.file;
-
-  console.log(chalk.blue('📋 Fetching model information...\n'));
-
-  try {
-    // Get model details
-    const modelDetails = await getModelDetails(parsed.repo);
-
-    // Display model information
-    displayModelInfo(modelDetails);
-
-    // If specific file requested, show file details
-    if (filename) {
-      console.log(chalk.blue('\n📄 File Details:\n'));
-      await displayFileInfo(parsed.repo, filename);
-    } else {
-      // Show all GGUF files
-      console.log(chalk.blue('\n📦 Available GGUF Files:\n'));
-      await displayAllFiles(parsed.repo);
-    }
-  } catch (error) {
-    throw new Error(`Failed to fetch model details: ${(error as Error).message}`);
-  }
-}
-
-async function getModelDetails(modelId: string): Promise<ModelDetails> {
-  return new Promise((resolve, reject) => {
-    const url = `https://huggingface.co/api/models/${modelId}`;
-
-    https.get(url, (response) => {
-      let data = '';
-
-      response.on('data', (chunk) => {
-        data += chunk;
-      });
-
-      response.on('end', () => {
-        try {
-          const json = JSON.parse(data);
-          const parts = modelId.split('/');
-
-          resolve({
-            modelId,
-            author: parts[0] || '',
-            modelName: parts.slice(1).join('/') || '',
-            downloads: json.downloads || 0,
-            likes: json.likes || 0,
-            lastModified: json.lastModified || '',
-            tags: json.tags || [],
-            library: json.library_name,
-            license: json.cardData?.license,
-          });
-        } catch (error) {
-          reject(new Error(`Failed to parse model data: ${(error as Error).message}`));
-        }
-      });
-    }).on('error', (error) => {
-      reject(error);
-    });
-  });
-}
-
-function displayModelInfo(details: ModelDetails): void {
-  console.log(chalk.bold('Model Information:'));
-  console.log(chalk.dim('─'.repeat(60)));
-  console.log(`${chalk.bold('ID:')}           ${details.modelId}`);
-  console.log(`${chalk.bold('Author:')}       ${details.author}`);
-  console.log(`${chalk.bold('Downloads:')}    ${details.downloads.toLocaleString()}`);
-  console.log(`${chalk.bold('Likes:')}        ${details.likes.toLocaleString()}`);
-  console.log(`${chalk.bold('Last Updated:')} ${new Date(details.lastModified).toLocaleDateString()}`);
-
-  if (details.license) {
-    console.log(`${chalk.bold('License:')}      ${details.license}`);
-  }
-
-  if (details.tags.length > 0) {
-    const relevantTags = details.tags
-      .filter(tag => !tag.startsWith('arxiv:') && !tag.startsWith('dataset:'))
-      .slice(0, 5);
-    if (relevantTags.length > 0) {
-      console.log(`${chalk.bold('Tags:')}         ${relevantTags.join(', ')}`);
-    }
-  }
-}
-
-async function displayFileInfo(modelId: string, filename: string): Promise<void> {
-  const files = await getModelFiles(modelId);
-  const file = files.find(f => f.filename === filename);
-
-  if (!file) {
-    console.log(chalk.yellow(`⚠️  File not found: ${filename}`));
-    console.log(chalk.dim('\nAvailable files:'));
-    files.forEach(f => console.log(chalk.dim(`  - ${f.filename}`)));
-    return;
-  }
-
-  const size = file.lfs?.size || file.size || 0;
-
-  console.log(`${chalk.bold('Filename:')}     ${file.filename}`);
-  console.log(`${chalk.bold('Size:')}         ${formatBytes(size)}`);
-
-  if (file.lfs) {
-    console.log(`${chalk.bold('SHA256:')}       ${file.lfs.oid.substring(0, 16)}...`);
-  }
-
-  console.log(chalk.dim('\nTo download:'));
-  console.log(chalk.dim(`  llamacpp pull ${modelId}/${filename}`));
-}
-
-async function displayAllFiles(modelId: string): Promise<void> {
-  const files = await getModelFiles(modelId);
-  const ggufFiles = files.filter(f => f.filename.toLowerCase().endsWith('.gguf'));
-
-  if (ggufFiles.length === 0) {
-    console.log(chalk.yellow('No GGUF files found in this model.'));
-    return;
-  }
-
-  const table = new Table({
-    head: ['FILENAME', 'SIZE'],
-    colWidths: [55, 12],
-  });
-
-  for (const file of ggufFiles) {
-    // LFS files store size in the lfs object, regular files in size field
-    const size = (file.lfs && typeof file.lfs.size === 'number') ? file.lfs.size : file.size;
-    table.push([
-      file.filename,
-      size > 0 ? formatBytes(size) : chalk.dim('Unknown'),
-    ]);
-  }
-
-  console.log(table.toString());
-
-  const totalSize = ggufFiles.reduce((sum, f) => {
-    const size = (f.lfs && typeof f.lfs.size === 'number') ? f.lfs.size : f.size;
-    return sum + size;
-  }, 0);
-  console.log(chalk.dim(`\nTotal: ${ggufFiles.length} files (${formatBytes(totalSize)})`));
-  console.log(chalk.dim('\nTo download a specific file:'));
-  console.log(chalk.dim(`  llamacpp pull ${modelId}/<filename>`));
-}
-
-async function getModelFiles(modelId: string): Promise<FileDetails[]> {
-  return new Promise((resolve, reject) => {
-    const url = `https://huggingface.co/api/models/${modelId}`;
-
-    https.get(url, (response) => {
-      let data = '';
-
-      response.on('data', (chunk) => {
-        data += chunk;
-      });
-
-      response.on('end', () => {
-        try {
-          const json = JSON.parse(data);
-          const files: FileDetails[] = (json.siblings || []).map((file: any) => ({
-            filename: file.rfilename,
-            size: file.size || 0,
-            lfs: file.lfs,
-          }));
-          resolve(files);
-        } catch (error) {
-          reject(new Error(`Failed to parse file data: ${(error as Error).message}`));
-        }
-      });
-    }).on('error', (error) => {
-      reject(error);
-    });
-  });
-}

package/src/commands/start.ts

@@ -1,101 +0,0 @@
-import chalk from 'chalk';
-import { stateManager } from '../lib/state-manager';
-import { launchctlManager } from '../lib/launchctl-manager';
-import { statusChecker } from '../lib/status-checker';
-import { parseMetalMemoryFromLog } from '../utils/file-utils';
-import { autoRotateIfNeeded, formatFileSize } from '../utils/log-utils';
-
-export async function startCommand(identifier: string): Promise<void> {
-  // Initialize state manager
-  await stateManager.initialize();
-
-  // 1. Find server by identifier
-  const server = await stateManager.findServer(identifier);
-  if (!server) {
-    throw new Error(
-      `Server not found: ${identifier}\n\n` +
-        `Use: llamacpp ps\n` +
-        `Or create a new server: llamacpp server create <model>`
-    );
-  }
-
-  // 2. Check if already running
-  if (server.status === 'running') {
-    console.log(
-      chalk.yellow(
-        `⚠️  Server ${server.modelName} is already running on port ${server.port}`
-      )
-    );
-    return;
-  }
-
-  console.log(chalk.blue(`▶️  Starting ${server.modelName} (port ${server.port})...`));
-
-  // 3. Auto-rotate logs if they exceed 100MB
-  try {
-    const result = await autoRotateIfNeeded(server.stdoutPath, server.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}`));
-  }
-
-  // 4. Ensure plist exists (recreate if missing)
-  try {
-    await launchctlManager.createPlist(server);
-  } catch (error) {
-    // May already exist, that's okay
-  }
-
-  // 5. Load service if needed
-  try {
-    await launchctlManager.loadService(server.plistPath);
-  } catch (error) {
-    // May already be loaded, that's okay
-  }
-
-  // 6. Start the service
-  try {
-    await launchctlManager.startService(server.label);
-  } catch (error) {
-    throw new Error(`Failed to start service: ${(error as Error).message}`);
-  }
-
-  // 7. Wait for startup
-  console.log(chalk.dim('Waiting for server to start...'));
-  const started = await launchctlManager.waitForServiceStart(server.label, 5000);
-
-  if (!started) {
-    throw new Error(
-      `Server failed to start. Check logs with: llamacpp server logs ${server.id}`
-    );
-  }
-
-  // 8. Update server status
-  let updatedServer = await statusChecker.updateServerStatus(server);
-
-  // 9. Parse Metal (GPU) memory allocation if not already captured
-  if (!updatedServer.metalMemoryMB) {
-    console.log(chalk.dim('Detecting Metal (GPU) memory allocation...'));
-    await new Promise(resolve => setTimeout(resolve, 8000)); // 8 second delay
-    const metalMemoryMB = await parseMetalMemoryFromLog(updatedServer.stderrPath);
-    if (metalMemoryMB) {
-      updatedServer = { ...updatedServer, metalMemoryMB };
-      await stateManager.saveServerConfig(updatedServer);
-      console.log(chalk.dim(`Metal memory: ${metalMemoryMB.toFixed(0)} MB`));
-    }
-  }
-
-  // 10. Display success
-  console.log();
-  console.log(chalk.green('✅ Server started successfully!'));
-  console.log();
-  console.log(chalk.dim(`Connect: http://localhost:${server.port}`));
-  console.log(chalk.dim(`View logs: llamacpp server logs ${server.id}`));
-  console.log(chalk.dim(`Stop: llamacpp server stop ${server.id}`));
-}

package/src/commands/stop.ts

@@ -1,39 +0,0 @@
-import chalk from 'chalk';
-import { stateManager } from '../lib/state-manager';
-import { launchctlManager } from '../lib/launchctl-manager';
-import { statusChecker } from '../lib/status-checker';
-
-export async function stopCommand(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`);
-  }
-
-  // Check if already stopped
-  if (server.status === 'stopped') {
-    console.log(chalk.yellow(`⚠️  Server ${server.modelName} is already stopped`));
-    return;
-  }
-
-  console.log(chalk.blue(`⏹️  Stopping ${server.modelName} (port ${server.port})...`));
-
-  // Unload the service (removes from launchd management - won't auto-restart)
-  try {
-    await launchctlManager.unloadService(server.plistPath);
-  } catch (error) {
-    throw new Error(`Failed to unload service: ${(error as Error).message}`);
-  }
-
-  // Wait for clean shutdown
-  const stopped = await launchctlManager.waitForServiceStop(server.label, 5000);
-
-  if (!stopped) {
-    console.log(chalk.yellow('⚠️  Server did not stop cleanly (timeout)'));
-  }
-
-  // Update server status
-  await statusChecker.updateServerStatus(server);
-
-  console.log(chalk.green('✅ Server stopped'));
-}

package/src/commands/tui.ts

@@ -1,25 +0,0 @@
-import chalk from 'chalk';
-import blessed from 'blessed';
-import { stateManager } from '../lib/state-manager.js';
-import { statusChecker } from '../lib/status-checker.js';
-import { createRootNavigator } from '../tui/RootNavigator.js';
-
-export async function tuiCommand(): Promise<void> {
-  const servers = await stateManager.getAllServers();
-
-  if (servers.length === 0) {
-    console.log(chalk.yellow('No servers configured.'));
-    console.log(chalk.dim('\nCreate a server: llamacpp server create <model-filename>'));
-    return;
-  }
-
-  const serversWithStatus = await statusChecker.updateAllServerStatuses();
-
-  const screen = blessed.screen({
-    smartCSR: true,
-    title: 'llama.cpp Server Monitor',
-    fullUnicode: true,
-  });
-
-  await createRootNavigator(screen, serversWithStatus);
-}

package/src/lib/config-generator.ts

@@ -1,130 +0,0 @@
-import * as os from 'os';
-import * as path from 'path';
-import { ServerConfig, sanitizeModelName } from '../types/server-config';
-import { getLogsDir, getLaunchAgentsDir } from '../utils/file-utils';
-import { stateManager } from './state-manager';
-
-export interface ServerOptions {
-  port?: number;
-  host?: string;
-  threads?: number;
-  ctxSize?: number;
-  gpuLayers?: number;
-  embeddings?: boolean;
-  jinja?: boolean;
-  verbose?: boolean;
-  customFlags?: string[];
-}
-
-export interface SmartDefaults {
-  threads: number;
-  ctxSize: number;
-  gpuLayers: number;
-}
-
-export class ConfigGenerator {
-  /**
-   * Calculate smart defaults based on model size
-   */
-  calculateSmartDefaults(modelSizeBytes: number): SmartDefaults {
-    const sizeGB = modelSizeBytes / (1024 ** 3);
-
-    // Context size based on model size
-    let ctxSize: number;
-    if (sizeGB < 1) {
-      ctxSize = 2048;        // < 1GB: small context
-    } else if (sizeGB < 3) {
-      ctxSize = 4096;        // 1-3GB: medium
-    } else if (sizeGB < 6) {
-      ctxSize = 8192;        // 3-6GB: large
-    } else {
-      ctxSize = 16384;       // 6GB+: very large
-    }
-
-    // GPU layers - always max for Metal (macOS)
-    const gpuLayers = 60;  // llama.cpp auto-detects optimal value
-
-    // Threads - use half of available cores (better performance)
-    const cpuCount = os.cpus().length;
-    const threads = Math.max(4, Math.floor(cpuCount / 2));
-
-    return { threads, ctxSize, gpuLayers };
-  }
-
-  /**
-   * Generate server configuration
-   */
-  async generateConfig(
-    modelPath: string,
-    modelName: string,
-    modelSize: number,
-    port: number,
-    options?: ServerOptions
-  ): Promise<ServerConfig> {
-    // Calculate smart defaults
-    const smartDefaults = this.calculateSmartDefaults(modelSize);
-
-    // Apply user overrides
-    const host = options?.host ?? '127.0.0.1';  // Default to localhost (secure)
-    const threads = options?.threads ?? smartDefaults.threads;
-    const ctxSize = options?.ctxSize ?? smartDefaults.ctxSize;
-    const gpuLayers = options?.gpuLayers ?? smartDefaults.gpuLayers;
-    const embeddings = options?.embeddings ?? true;
-    const jinja = options?.jinja ?? true;
-    const verbose = options?.verbose ?? true;  // Default to true (HTTP request logging)
-    const customFlags = options?.customFlags;  // Optional custom flags
-
-    // Generate server ID
-    const id = sanitizeModelName(modelName);
-
-    // Generate paths
-    const label = `com.llama.${id}`;
-    const plistPath = path.join(getLaunchAgentsDir(), `${label}.plist`);
-    const logsDir = getLogsDir();
-    const stdoutPath = path.join(logsDir, `${id}.stdout`);
-    const stderrPath = path.join(logsDir, `${id}.stderr`);
-
-    const config: ServerConfig = {
-      id,
-      modelPath,
-      modelName,
-      port,
-      host,
-      threads,
-      ctxSize,
-      gpuLayers,
-      embeddings,
-      jinja,
-      verbose,
-      customFlags,
-      status: 'stopped',
-      createdAt: new Date().toISOString(),
-      plistPath,
-      label,
-      stdoutPath,
-      stderrPath,
-    };
-
-    return config;
-  }
-
-  /**
-   * Merge global defaults with user options
-   */
-  async mergeWithGlobalDefaults(options?: ServerOptions): Promise<Partial<ServerOptions>> {
-    const globalConfig = await stateManager.loadGlobalConfig();
-
-    return {
-      host: options?.host ?? '127.0.0.1',
-      threads: options?.threads ?? globalConfig.defaults.threads,
-      ctxSize: options?.ctxSize ?? globalConfig.defaults.ctxSize,
-      gpuLayers: options?.gpuLayers ?? globalConfig.defaults.gpuLayers,
-      embeddings: options?.embeddings ?? true,
-      jinja: options?.jinja ?? true,
-      verbose: options?.verbose ?? true,
-    };
-  }
-}
-
-// Export singleton instance
-export const configGenerator = new ConfigGenerator();