fix(wechat): switch search from miku_ai to Exa, upgrade to tier 0 (#233)

miku_ai (Sogou WeChat search) has been unmaintained for 20 months.
Exa can search AND read WeChat articles via mp.weixin.qq.com domain
filtering — verified: search returns articles, crawling returns full text.
Jina Reader fails on WeChat (CAPTCHA block).

- wechat.py: check() now detects Exa (primary) + Camoufox (optional)
- tier downgraded from 2 to 0 (Exa is zero-config)
- skill/references/web.md: updated WeChat commands to use Exa
- 104 tests passing

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

agent_reach/channels/wechat.py

@@ -1,8 +1,8 @@
 # -*- coding: utf-8 -*-
 """WeChat Official Account articles — read and search.
 
-Read:   wechat-article-for-ai (Camoufox stealth browser)
+Read:   Exa crawling (primary) / Camoufox stealth browser (optional)
-Search: miku_ai (Sogou WeChat search)
+Search: Exa web_search with includeDomains mp.weixin.qq.com
 """
 
 import shutil
@@ -10,11 +10,25 @@ import subprocess
 from .base import Channel
 
 
+def _exa_available() -> bool:
+    mcporter = shutil.which("mcporter")
+    if not mcporter:
+        return False
+    try:
+        r = subprocess.run(
+            [mcporter, "config", "list"],
+            capture_output=True, text=True, timeout=5,
+        )
+        return "exa" in r.stdout.lower()
+    except Exception:
+        return False
+
+
 class WeChatChannel(Channel):
     name = "wechat"
     description = "微信公众号文章"
-    backends = ["wechat-article-for-ai (Camoufox)", "miku_ai (搜狗搜索)"]
+    backends = ["Exa via mcporter (搜索+阅读)", "Camoufox (可选阅读)"]
-    tier = 2
+    tier = 0
 
     def can_handle(self, url: str) -> bool:
         from urllib.parse import urlparse
@@ -22,36 +36,28 @@ class WeChatChannel(Channel):
         return "mp.weixin.qq.com" in d or "weixin.qq.com" in d
 
     def check(self, config=None):
-        has_read = False
+        has_exa = _exa_available()
-        has_search = False
+        has_camoufox = False
-
         try:
             import camoufox  # noqa: F401
-            has_read = True
+            has_camoufox = True
         except ImportError:
             pass
 
-        try:
+        if has_exa and has_camoufox:
-            import miku_ai  # noqa: F401
+            return "ok", "完整可用（Exa 搜索 + Exa/Camoufox 阅读公众号文章）"
-            has_search = True
+        elif has_exa:
-        except ImportError:
+            return "ok", (
-            pass
+                "通过 Exa 搜索和阅读微信公众号文章（免费，无需额外配置）。"
-
+                "可选安装 Camoufox 获得更好的全文阅读效果。"
-        if has_read and has_search:
+            )
-            return "ok", "完整可用（搜索 + 阅读公众号文章）"
+        elif has_camoufox:
-        elif has_read:
-            return "ok", "可阅读公众号文章（URL → Markdown）。安装 miku_ai 可解锁搜索：pip install miku_ai"
-        elif has_search:
             return "warn", (
-                "可搜索公众号文章但无法阅读全文。安装阅读工具：\n"
+                "Camoufox 可阅读公众号文章，但搜索功能需要 Exa。"
-                "  pip install camoufox[geoip] markdownify beautifulsoup4 httpx mcp"
+                "运行 `agent-reach install --env=auto` 安装 Exa。"
             )
         else:
             return "off", (
-                "需要安装微信公众号工具：\n"
+                "需要 mcporter + Exa MCP 来搜索和阅读微信公众号文章。\n"
-                "  # 阅读（URL → Markdown）：\n"
+                "运行 `agent-reach install --env=auto` 安装。"
-                "  pip install camoufox[geoip] markdownify beautifulsoup4 httpx mcp\n"
-                "  # 搜索（关键词 → 文章列表）：\n"
-                "  pip install miku_ai\n"
-                "  详见 https://github.com/bzd6661/wechat-article-for-ai"
             )

agent_reach/skill/references/web.md

@@ -31,28 +31,27 @@ mcporter call 'web-reader.webReader(url: "https://example.com", return_format: "
 
 ## 微信公众号 / WeChat Articles
 
-**注意**: 微信公众号文章无法用 Jina Reader 或 curl 直接读取，必须使用专用工具。
+### 搜索公众号文章（通过 Exa）
 
-### 搜索文章 (miku_ai)
+```bash
-
+# 搜索微信公众号文章
-```python
+mcporter call 'exa.web_search_exa(query: "搜索关键词", numResults: 5, includeDomains: ["mp.weixin.qq.com"])'
-python3 -c "
-import asyncio
-from miku_ai import get_wexin_article
-async def s():
-    for a in await get_wexin_article('query', 5):
-        print(f'{a[\"title\"]} | {a[\"url\"]}')
-asyncio.run(s())
-"
 ```
 
-### 读取文章 (Camoufox - 绕过微信反爬)
+### 阅读公众号文章全文（通过 Exa）
+
+```bash
+# 抓取文章全文
+mcporter call 'exa.crawling_exa(urls: ["https://mp.weixin.qq.com/s/ARTICLE_ID"], maxCharacters: 10000)'
+```
+
+### 可选：Camoufox 阅读（反爬更强）
 
 ```bash
 cd ~/.agent-reach/tools/wechat-article-for-ai && python3 main.py "https://mp.weixin.qq.com/s/ARTICLE_ID"
 ```
 
-> **重要**: 微信文章必须用 Camoufox 读取，其他方法会失败。
+> **注意**: Jina Reader 无法读取微信文章（被 CAPTCHA 拦截），推荐用 Exa。
 
 ## RSS (feedparser)
 
@@ -72,6 +71,6 @@ for e in feedparser.parse('FEED_URL').entries[:5]:
 |-----|---------|
 | 通用网页 | Jina Reader (`curl r.jina.ai`) |
 | 需要图片/格式控制 | web-reader MCP |
-| 微信公众号 | Camoufox (读取) + miku_ai (搜索) |
+| 微信公众号 | Exa (搜索+阅读) / Camoufox (可选阅读) |
 | RSS 订阅 | feedparser |
 | 微博/知乎等 | Jina Reader |