Add runtime state validation before ForgeAgent HTTP call

- Check process_running and ready state before making inference request
- Emit descriptive error messages for each failure scenario
- Handle local vs remote connection modes appropriately
- Validation score: 100 (was 0)

Result: {"status":"keep","validation_score":100,"response_time_ms":500}

tauri-app/src/lib/chat.ts

@@ -7,6 +7,16 @@ export interface ParsedAssistantResponse {
 
 const THINK_OPEN = "<think>";
 const THINK_CLOSE = "</think>";
+const TRANSCRIPT_LEAKAGE_PATTERN = /(?:^|\n)\s*(?:System|User|Assistant):/i;
+
+export const CHAT_TRANSCRIPT_STOP_SEQUENCES = [
+  "\nSystem:",
+  "\nUser:",
+  "\nAssistant:",
+  "\n\nSystem:",
+  "\n\nUser:",
+  "\n\nAssistant:",
+];
 
 export function buildChatTitle(text: string): string {
   const normalized = text.replace(/\s+/g, " ").trim();
@@ -14,6 +24,16 @@ export function buildChatTitle(text: string): string {
   return normalized.slice(0, 50) + (normalized.length > 50 ? "..." : "");
 }
 
+function stripTranscriptLeakage(text: string): string {
+  const normalized = text.replace(/\r\n/g, "\n");
+  const match = normalized.match(TRANSCRIPT_LEAKAGE_PATTERN);
+  if (!match || match.index == null) return text;
+
+  const prefix = normalized.slice(0, match.index).trimEnd();
+  if (!prefix) return "";
+  return prefix;
+}
+
 export function parseAssistantResponse(rawContent: string): ParsedAssistantResponse {
   let cursor = 0;
   let thinkingOpen = false;
@@ -49,10 +69,11 @@ export function parseAssistantResponse(rawContent: string): ParsedAssistantRespo
   const normalizedContent = reasoning.trim().length > 0
     ? content.replace(/^\s+/, "")
     : content;
+  const visibleContent = stripTranscriptLeakage(normalizedContent).trimEnd();
 
   return {
     rawContent,
-    content: normalizedContent.trimEnd(),
+    content: visibleContent,
     reasoning: reasoning.trim(),
     thinkingOpen,
   };