fix(cli): EADDRINUSE detection via PID file fallback (#3388)

Closes #3346. isServerHealthy() falls back to PID file check when /v1/health times out. Clear EADDRINUSE error message with ag stop suggestion.

Author: OneStepAt4time

src/commands/run.ts

@@ -62,13 +62,43 @@ async function verifyAuth(baseUrl: string, authToken: string | undefined): Promi
   }
 }
 
-/** Check if the server is healthy at the given base URL. */
+/** Check if the server is healthy at the given base URL.
+ *  #3346: Also checks PID file as a fallback when the health endpoint times out.
+ *  If the PID file exists and the process is alive, assume the server is running. */
 async function isServerHealthy(baseUrl: string, authToken?: string): Promise<boolean> {
   try {
     const headers: Record<string, string> = {};
     if (authToken) headers['Authorization'] = `Bearer ${authToken}`;
     const res = await fetch(`${baseUrl}/v1/health`, { headers, signal: AbortSignal.timeout(3000) });
-    return res.ok;
+    if (res.ok) return true;
+    // Non-401 error or network issue — fall through to PID file check
+  } catch {
+    // Network error — fall through to PID file check
+  }
+
+  // #3346: Fallback — check if an Aegis server PID file exists with a live process
+  try {
+    const { existsSync, readFileSync } = await import('node:fs');
+    const stateDir = process.env.AEGIS_STATE_DIR || join(homedir(), '.aegis');
+    const pidFile = join(stateDir, 'aegis.pid');
+    if (existsSync(pidFile)) {
+      const pid = parseInt(readFileSync(pidFile, 'utf-8').trim(), 10);
+      if (pid > 0 && pidExists(pid)) {
+        return true;
+      }
+    }
+  } catch {
+    // Ignore
+  }
+
+  return false;
+}
+
+/** Check if a process with the given PID is alive. */
+function pidExists(pid: number): boolean {
+  try {
+    process.kill(pid, 0);
+    return true;
   } catch {
     return false;
   }