Fix AI output showing UTF-8 bytes as hex escapes instead of Chinese.

Decode <0xE5><0xA7><0xA4> style model output to proper characters; add prompt rule to use normal Chinese text.

Co-authored-by: Cursor <cursoragent@cursor.com>

lib/ai/decode-text.ts

@@ -0,0 +1,20 @@
+/**
+ * 部分模型会把 UTF-8 汉字输出成 <0xE5><0xA7><0xA4> 形式，还原为正常文字。
+ * 例：<0xE5><0xA7><0xA4> → 姤
+ */
+export function decodeHexByteEscapes(text: string): string {
+  return text.replace(/(?:<0x[0-9A-Fa-f]{2}>)+/gi, (match) => {
+    const parts = match.match(/<0x([0-9A-Fa-f]{2})>/gi);
+    if (!parts?.length) {
+      return match;
+    }
+    const bytes = Uint8Array.from(
+      parts.map((part) => parseInt(part.slice(3, 5), 16)),
+    );
+    try {
+      return new TextDecoder("utf-8", { fatal: true }).decode(bytes);
+    } catch {
+      return new TextDecoder("utf-8", { fatal: false }).decode(bytes);
+    }
+  });
+}