feat(automation): complete unified automation system redesign

Phase 4 completion:
- Add ApprovalQueue component for managing pending approvals
- Add ExecutionResult component for displaying hand/workflow results
- Update Sidebar navigation to use unified AutomationPanel
- Replace separate 'hands' and 'workflow' tabs with single 'automation' tab
- Fix TypeScript type safety issues with unknown types in JSX expressions

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

desktop/src/lib/llm-service.ts

@@ -295,21 +295,59 @@ class GatewayLLMAdapter implements LLMServiceAdapter {
     const config = { ...this.config, ...options };
     const startTime = Date.now();
 
-    const response = await fetch(`${config.apiBase}/complete`, {
+    // Build a single prompt from messages
+    const systemMessage = messages.find(m => m.role === 'system')?.content || '';
+    const userMessage = messages.find(m => m.role === 'user')?.content || '';
+
+    // Combine system and user messages into a single prompt
+    const fullPrompt = systemMessage
+      ? `${systemMessage}\n\n${userMessage}`
+      : userMessage;
+
+    // Use OpenFang's chat endpoint (same as main chat)
+    // Try to get the default agent ID from localStorage or use 'default'
+    const agentId = localStorage.getItem('zclaw-default-agent-id') || 'default';
+
+    const response = await fetch(`/api/agents/${agentId}/message`, {
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
       },
       body: JSON.stringify({
-        messages,
+        message: fullPrompt,
         max_tokens: config.maxTokens,
-        temperature: config.temperature,
+        temperature: config.temperature ?? 0.3, // Lower temperature for extraction tasks
       }),
       signal: AbortSignal.timeout(config.timeout || 60000),
     });
 
     if (!response.ok) {
       const error = await response.text();
+      // If agent not found, try without agent ID (direct /api/chat)
+      if (response.status === 404) {
+        const fallbackResponse = await fetch('/api/chat', {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({
+            message: fullPrompt,
+            max_tokens: config.maxTokens,
+            temperature: config.temperature ?? 0.3,
+          }),
+          signal: AbortSignal.timeout(config.timeout || 60000),
+        });
+
+        if (!fallbackResponse.ok) {
+          throw new Error(`[Gateway] Both endpoints failed: ${fallbackResponse.status}`);
+        }
+
+        const data = await fallbackResponse.json();
+        const latencyMs = Date.now() - startTime;
+        return {
+          content: data.response || data.content || '',
+          tokensUsed: { input: data.input_tokens || 0, output: data.output_tokens || 0 },
+          latencyMs,
+        };
+      }
       throw new Error(`[Gateway] API error: ${response.status} - ${error}`);
     }
 
@@ -317,15 +355,14 @@ class GatewayLLMAdapter implements LLMServiceAdapter {
     const latencyMs = Date.now() - startTime;
 
     return {
-      content: data.content || data.choices?.[0]?.message?.content || '',
-      tokensUsed: data.tokensUsed || { input: 0, output: 0 },
-      model: data.model,
+      content: data.response || data.content || '',
+      tokensUsed: { input: data.input_tokens || 0, output: data.output_tokens || 0 },
       latencyMs,
     };
   }
 
   isAvailable(): boolean {
-    // Gateway is available if we're connected to OpenFang
+    // Gateway is available if we're in browser (can connect to OpenFang)
     return typeof window !== 'undefined';
   }
 
@@ -382,8 +419,8 @@ export function loadConfig(): LLMConfig {
     // Ignore parse errors
   }
 
-  // Default to mock for safety
-  return DEFAULT_CONFIGS.mock;
+  // Default to gateway (OpenFang passthrough) for L4 self-evolution
+  return DEFAULT_CONFIGS.gateway;
 }
 
 export function saveConfig(config: LLMConfig): void {