feat(v1.8.1): Windows 兼容性修复 + SKILL.md 铁律区块

Windows 兼容性修复（基于真实使用痛点）:

- force_utf8_stdout(): 强制 stdout/stderr 为 UTF-8，修复 Windows GBK 崩溃（print('\\xa0') 不再炸）

- resolve_instances/load_config 新增 %APPDATA%/searxng-cli/ 路径，覆盖 Windows 配置约定

- fetch.py 失败诊断增强：输出 status_code=/cause=/url= 字段，AI Agent 可程序化区分 404/403/DNS 失败

SKILL.md 铁律区块（5 条，置顶）:

- stdout=数据/stderr=日志 永不混淆

- 禁用 2>/dev/null（丢弃 stderr = 失败时零诊断）

- 排错去 --quiet 加 --verbose

- 配置查找覆盖 WSL + Windows 双路径

- 实例 URL 必填，公共实例发现已移除

测试: 352 -> 362（新增 10 个：force_utf8_stdout 幂等性/GBK 替换/非 ASCII 打印/APPDATA 路径发现/txt 回退/空 APPDATA）

scripts/common.py

@@ -11,12 +11,14 @@ This module centralizes code that was previously duplicated across
 * ``FALLBACK_UAS``        — browser-like User-Agents used when blocked
 * retry constants         — ``RETRYABLE_STATUS``, ``RETRY_BACKOFF_BASE``, etc.
 * ``setup_logging``       — shared logging configuration (--verbose/--quiet)
+* ``force_utf8_stdout``   — force stdout to UTF-8 (fix Windows GBK crashes)
 
 Centralizing the retry policy guarantees that both scripts treat 429/5xx
 as retryable and connection errors as transient, eliminating the previous
 inconsistency where ``search.py`` ignored 5xx.
 """
 
+import io
 import logging
 import sys
 import urllib.error
@@ -56,6 +58,58 @@ def setup_logging(verbose: bool = False, quiet: bool = False) -> None:
     # Don't let root logger add its own handler — we own the searxng namespace.
     _LOG.propagate = False
 
+
+def force_utf8_stdout() -> None:
+    """Force stdout/stderr to UTF-8 to prevent Windows GBK encoding crashes.
+
+    Windows Python defaults ``sys.stdout`` to the OEM codepage (often GBK on
+    Chinese Windows). ``print()`` of any character outside that codepage
+    (e.g. ``\\xa0`` nbsp, CJK punctuation from foreign pages) raises
+    ``UnicodeEncodeError`` and kills the process.
+
+    ``PYTHONIOENCODING=utf-8`` is unreliable here because Python 3.7+
+    reconfigures stdout after reading that env var in some scenarios (e.g.
+    when stdout has already been wrapped). The only reliable fix is to
+    reconfigure the stream in-process.
+
+    Uses ``sys.stdout.reconfigure()`` on Python 3.7+, falling back to
+    wrapping ``sys.stdout.buffer`` on older versions. Both paths use
+    ``errors='replace'`` so an undecodable byte never crashes the script —
+    better to emit ``?`` than to lose all output.
+
+    Safe to call multiple times; subsequent calls are no-ops once the
+    encoding is already UTF-8 (or close enough — we check the lowercased
+    encoding name to tolerate ``utf-8`` vs ``UTF-8`` vs ``utf8``).
+    """
+    for stream_name in ("stdout", "stderr"):
+        stream = getattr(sys, stream_name, None)
+        if stream is None:
+            continue
+        # Already UTF-8? Skip (covers Linux/macOS and re-invoked scripts).
+        enc = getattr(stream, "encoding", "") or ""
+        if enc.lower().replace("-", "") in ("utf8", "utf-8-sig"):
+            continue
+        # Python 3.7+ has TextIOWrapper.reconfigure()
+        reconfigure = getattr(stream, "reconfigure", None)
+        if reconfigure is not None:
+            try:
+                reconfigure(encoding="utf-8", errors="replace")
+                continue
+            except (ValueError, OSError):
+                pass  # Fall through to the buffer-wrap path
+        # Fallback: wrap the underlying buffer in a new UTF-8 stream.
+        buffer = getattr(stream, "buffer", None)
+        if buffer is not None:
+            try:
+                new_stream = io.TextIOWrapper(
+                    buffer, encoding="utf-8", errors="replace", line_buffering=True,
+                )
+                setattr(sys, stream_name, new_stream)
+            except (ValueError, AttributeError):
+                # Last resort: keep the original stream. Better to risk a
+                # GBK crash on exotic characters than to break stdout entirely.
+                pass
+
 # Retry settings (shared by both scripts)
 MAX_RETRIES = 3
 RETRY_BACKOFF_BASE = 1.5  # seconds; exponential backoff + jitter