feat: Z.AI provider integration, World Model view, enhanced model support

README.md

@@ -73,9 +73,9 @@ Traditional development tools help you write code. Automaker helps you **orchest
 4. **Review & Verify** - Review the changes, run tests, and approve when ready
 5. **Ship Faster** - Build entire applications in days, not weeks
 
-### Powered by Claude Agent SDK
+### Powered by Claude Agent SDK & Z.AI
 
-Automaker leverages the [Claude Agent SDK](https://www.npmjs.com/package/@anthropic-ai/claude-agent-sdk) to give AI agents full access to your codebase. Agents can read files, write code, execute commands, run tests, and make git commits—all while working in isolated git worktrees to keep your main branch safe. The SDK provides autonomous AI agents that can use tools, make decisions, and complete complex multi-step tasks without constant human intervention.
+Automaker leverages the [Claude Agent SDK](https://www.npmjs.com/package/@anthropic-ai/claude-agent-sdk) and proprietary **Z.AI Provider** to give AI agents full access to your codebase. Configure agents to use Anthropic's Claude 3.5 Sonnet or **Z.AI's GLM-4** models for powerful, reasoning-first development. Agents can read files, write code, execute commands, run tests, and make git commits—all while working in isolated git worktrees to keep your main branch safe. The SDK provides autonomous AI agents that can use tools, make decisions, and complete complex multi-step tasks without constant human intervention.
 
 ### Why This Matters
 
@@ -130,7 +130,8 @@ npm run dev
 **Authentication Setup:** On first run, Automaker will automatically show a setup wizard where you can configure authentication. You can choose to:
 
 - Use **Claude Code CLI** (recommended) - Automaker will detect your CLI credentials automatically
-- Enter an **API key** directly in the wizard
+- Enter an **Anthropic API key** directly in the wizard
+- Enter a **Z.AI API key** directly in the wizard (for Z.AI models)
 
 If you prefer to set up authentication before running (e.g., for headless deployments or CI/CD), you can set it manually:
 
@@ -412,7 +413,7 @@ The application can store your API key securely in the settings UI. The key is p
 
 ### AI & Planning
 
-- 🧠 **Multi-Model Support** - Choose from Claude Opus, Sonnet, and Haiku per feature
+- 🧠 **Multi-Model Support** - Choose from Claude Opus, Sonnet, Haiku, or **Z.AI GLM-4** per feature
 - 💭 **Extended Thinking** - Enable thinking modes (none, medium, deep, ultra) for complex problem-solving
 - 📝 **Planning Modes** - Four planning levels: skip (direct implementation), lite (quick plan), spec (task breakdown), full (phased execution)
 - ✅ **Plan Approval** - Review and approve AI-generated plans before implementation begins
@@ -477,6 +478,7 @@ The application can store your API key securely in the settings UI. The key is p
 - **Express 5** - HTTP server framework
 - **TypeScript 5.9** - Type safety
 - **Claude Agent SDK** - AI agent integration (@anthropic-ai/claude-agent-sdk)
+- **Z.AI Provider** - Custom integration for Z.AI models
 - **WebSocket (ws)** - Real-time event streaming
 - **node-pty** - PTY terminal sessions
 

apps/server/package.json

@@ -24,20 +24,44 @@
     "test:unit": "vitest run tests/unit"
   },
   "dependencies": {
+    "@ai-sdk/amazon-bedrock": "^4.0.16",
+    "@ai-sdk/anthropic": "^3.0.12",
+    "@ai-sdk/azure": "^3.0.9",
+    "@ai-sdk/baseten": "^1.0.8",
+    "@ai-sdk/cerebras": "^2.0.8",
+    "@ai-sdk/cohere": "^3.0.5",
+    "@ai-sdk/deepinfra": "^2.0.7",
+    "@ai-sdk/deepseek": "^2.0.5",
+    "@ai-sdk/fal": "^2.0.6",
+    "@ai-sdk/fireworks": "^2.0.7",
+    "@ai-sdk/google": "^3.0.7",
+    "@ai-sdk/google-vertex": "^4.0.14",
+    "@ai-sdk/groq": "^3.0.5",
+    "@ai-sdk/luma": "^2.0.5",
+    "@ai-sdk/mistral": "^3.0.6",
+    "@ai-sdk/openai": "^3.0.9",
+    "@ai-sdk/perplexity": "^3.0.5",
+    "@ai-sdk/togetherai": "^2.0.7",
+    "@ai-sdk/xai": "^3.0.20",
     "@anthropic-ai/claude-agent-sdk": "0.1.76",
     "@automaker/dependency-resolver": "1.0.0",
     "@automaker/git-utils": "1.0.0",
     "@automaker/model-resolver": "1.0.0",
     "@automaker/platform": "1.0.0",
     "@automaker/prompts": "1.0.0",
+    "@automaker/provider-claude": "1.0.0",
+    "@automaker/provider-zai": "1.0.0",
+    "@automaker/providers-core": "1.0.0",
     "@automaker/types": "1.0.0",
     "@automaker/utils": "1.0.0",
-    "@modelcontextprotocol/sdk": "1.25.1",
+    "@modelcontextprotocol/sdk": "^1.25.2",
     "@openai/codex-sdk": "^0.77.0",
+    "ai": "^6.0.31",
     "cookie-parser": "1.4.7",
     "cors": "2.8.5",
     "dotenv": "17.2.3",
     "express": "5.2.1",
+    "jsonwebtoken": "^9.0.3",
     "morgan": "1.10.1",
     "node-pty": "1.1.0-beta41",
     "ws": "8.18.3"
@@ -47,6 +71,7 @@
     "@types/cookie-parser": "1.4.10",
     "@types/cors": "2.8.19",
     "@types/express": "5.0.6",
+    "@types/jsonwebtoken": "^9.0.10",
     "@types/morgan": "1.9.10",
     "@types/node": "22.19.3",
     "@types/ws": "8.18.1",

apps/server/src/cluster-manager.ts

@@ -0,0 +1,153 @@
+/**
+ * Cluster Manager for AutoMaker Server
+ *
+ * Enables multi-core CPU utilization by spawning worker processes.
+ * The master process manages workers and restarts them on crash.
+ * Workers share the port via OS-level load balancing.
+ *
+ * Usage:
+ *   import { initCluster } from './cluster-manager.js';
+ *   initCluster(() => { startServer(); });
+ */
+
+import cluster from 'cluster';
+import os from 'os';
+import { createLogger } from './utils/logger.js';
+
+const logger = createLogger('Cluster');
+
+// Configuration
+const CLUSTER_ENABLED = process.env.CLUSTER_MODE === 'true';
+const WORKER_COUNT = parseInt(process.env.WORKER_COUNT || '0', 10) || os.cpus().length;
+const RESTART_DELAY_MS = 1000;
+
+// Track worker restarts to prevent rapid restart loops
+const workerRestarts = new Map<number, number>();
+const MAX_RESTARTS_PER_MINUTE = 5;
+
+/**
+ * Initialize cluster mode if enabled.
+ * In cluster mode, the master spawns workers that each run the server.
+ *
+ * @param startWorker - Function to start the server (called in each worker)
+ * @returns true if this process should continue (worker or non-cluster), false if master
+ */
+export function initCluster(startWorker: () => void): boolean {
+  // Skip cluster mode if disabled or in development
+  if (!CLUSTER_ENABLED) {
+    logger.info('Cluster mode disabled, running single-process');
+    startWorker();
+    return true;
+  }
+
+  if (cluster.isPrimary) {
+    logger.info(`Master process ${process.pid} starting ${WORKER_COUNT} workers`);
+
+    // Fork workers
+    for (let i = 0; i < WORKER_COUNT; i++) {
+      forkWorker();
+    }
+
+    // Handle worker exit - restart with exponential backoff
+    cluster.on('exit', (worker, code, signal) => {
+      const workerId = worker.id;
+      const exitReason = signal ? `signal ${signal}` : `code ${code}`;
+
+      if (code !== 0) {
+        logger.warn(`Worker ${workerId} (PID ${worker.process.pid}) died (${exitReason})`);
+
+        // Check restart rate limiting
+        const now = Date.now();
+        const lastRestarts = workerRestarts.get(workerId) || 0;
+
+        if (lastRestarts >= MAX_RESTARTS_PER_MINUTE) {
+          logger.error(`Worker ${workerId} restarted too many times, not restarting`);
+          return;
+        }
+
+        workerRestarts.set(workerId, lastRestarts + 1);
+
+        // Clear restart count after 1 minute
+        setTimeout(() => {
+          workerRestarts.set(workerId, Math.max(0, (workerRestarts.get(workerId) || 0) - 1));
+        }, 60000);
+
+        // Restart with delay
+        setTimeout(() => {
+          logger.info(`Restarting worker ${workerId}...`);
+          forkWorker();
+        }, RESTART_DELAY_MS);
+      } else {
+        logger.info(`Worker ${workerId} exited gracefully`);
+      }
+    });
+
+    // Handle graceful shutdown
+    const shutdown = () => {
+      logger.info('Master shutting down, terminating workers...');
+      for (const id in cluster.workers) {
+        cluster.workers[id]?.kill('SIGTERM');
+      }
+      process.exit(0);
+    };
+
+    process.on('SIGINT', shutdown);
+    process.on('SIGTERM', shutdown);
+
+    return false; // Master doesn't run server code
+  } else {
+    // Worker process - run the server
+    logger.info(`Worker ${cluster.worker?.id} (PID ${process.pid}) started`);
+    startWorker();
+    return true;
+  }
+}
+
+/**
+ * Fork a new worker process
+ */
+function forkWorker(): void {
+  const worker = cluster.fork();
+
+  // Handle worker messages (for inter-process communication if needed)
+  worker.on('message', (msg: { type: string; data?: unknown }) => {
+    if (msg.type === 'broadcast') {
+      // Broadcast message to all workers
+      for (const id in cluster.workers) {
+        if (cluster.workers[id] !== worker) {
+          cluster.workers[id]?.send(msg);
+        }
+      }
+    }
+  });
+}
+
+/**
+ * Check if this process is the master/primary
+ */
+export function isMaster(): boolean {
+  return cluster.isPrimary;
+}
+
+/**
+ * Get the current worker ID (0 if not in cluster mode or if master)
+ */
+export function getWorkerId(): number {
+  return cluster.worker?.id || 0;
+}
+
+/**
+ * Get total worker count
+ */
+export function getWorkerCount(): number {
+  return CLUSTER_ENABLED ? WORKER_COUNT : 1;
+}
+
+/**
+ * Broadcast a message to all workers (call from any worker)
+ */
+export function broadcastToWorkers(data: unknown): void {
+  if (cluster.isWorker && process.send) {
+    process.send({ type: 'broadcast', data });
+  }
+}

apps/server/src/diagnostics/test-write-persistence.ts

@@ -0,0 +1,71 @@
+/**
+ * Diagnostic Test: ZaiTools Write Persistence
+ *
+ * Tests whether ZaiTools.executeWriteFile correctly writes to disk.
+ * Run with: npx tsx apps/server/src/diagnostics/test-write-persistence.ts
+ */
+
+import path from 'path';
+import fs from 'fs/promises';
+import { ZaiTools } from '../providers/zai-tools.js';
+
+async function testWritePersistence() {
+  const testDir = process.cwd();
+  const testFilePath = path.join(testDir, 'test-write-persistence-output.txt');
+  const testContent = `Test file created at ${new Date().toISOString()}\nIf you see this, ZaiTools.Write is working correctly.`;
+
+  console.log('=== ZaiTools Write Persistence Test ===\n');
+  console.log(`Working directory: ${testDir}`);
+  console.log(`Test file path: ${testFilePath}\n`);
+
+  const zaiTools = new ZaiTools(testDir);
+
+  // Test 1: Write via executeTool
+  console.log('1. Testing executeTool("Write", {...})...');
+  const result = await zaiTools.executeTool('Write', {
+    path: testFilePath,
+    content: testContent,
+  });
+  console.log(`   Result: ${result}`);
+
+  // Test 2: Verify file exists
+  console.log('\n2. Verifying file exists on disk...');
+  try {
+    const stat = await fs.stat(testFilePath);
+    console.log(`   ✅ File exists! Size: ${stat.size} bytes`);
+  } catch (error) {
+    console.log(`   ❌ File does NOT exist: ${(error as Error).message}`);
+    console.log('   This confirms the Write tool is BROKEN.');
+    process.exit(1);
+  }
+
+  // Test 3: Read content back
+  console.log('\n3. Reading content back...');
+  try {
+    const readContent = await fs.readFile(testFilePath, 'utf-8');
+    if (readContent === testContent) {
+      console.log('   ✅ Content matches exactly!');
+    } else {
+      console.log('   ⚠️ Content differs:');
+      console.log(`   Expected: ${testContent.substring(0, 50)}...`);
+      console.log(`   Got: ${readContent.substring(0, 50)}...`);
+    }
+  } catch (error) {
+    console.log(`   ❌ Failed to read: ${(error as Error).message}`);
+  }
+
+  // Cleanup
+  console.log('\n4. Cleaning up test file...');
+  try {
+    await fs.unlink(testFilePath);
+    console.log('   ✅ Test file removed.');
+  } catch {
+    console.log('   ⚠️ Could not remove test file (non-critical).');
+  }
+
+  console.log('\n=== Test Complete ===');
+  console.log('If you see ✅ for steps 1-3, ZaiTools.Write is working correctly.');
+  console.log('The issue must be in how auto-mode-service invokes it during feature execution.\n');
+}
+
+testWritePersistence().catch(console.error);