From 476935be92c4e0a8f468e87585ecfc92ae44b439 Mon Sep 17 00:00:00 2001
From: Pnant <73925474+Panniantong@users.noreply.github.com>
Date: Wed, 4 Mar 2026 18:02:05 +0800
Subject: [PATCH] feat: add WeChat search via miku_ai (sogou) (#68)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- wechat channel now supports both search and read
- Search: miku_ai (sogou weixin search) → article list with URLs
- Read: camoufox (stealth Firefox) → full Markdown content
- Doctor shows granular status (search only / read only / both)
- SKILL.md updated with search + read workflow examples

Co-authored-by: Panniantong <panniantong@users.noreply.github.com>
---
 agent_reach/channels/wechat.py | 39 ++++++++++++++++++++++++++++------
 agent_reach/skill/SKILL.md     | 27 ++++++++++++++++++-----
 2 files changed, 54 insertions(+), 12 deletions(-)

diff --git a/agent_reach/channels/wechat.py b/agent_reach/channels/wechat.py
index 369d016..25df67a 100644
--- a/agent_reach/channels/wechat.py
+++ b/agent_reach/channels/wechat.py
@@ -1,5 +1,9 @@
 # -*- coding: utf-8 -*-
-"""WeChat Official Account articles — check if wechat-article-for-ai is available."""
+"""WeChat Official Account articles — read and search.
+
+Read:   wechat-article-for-ai (Camoufox stealth browser)
+Search: miku_ai (Sogou WeChat search)
+"""
 
 import shutil
 import subprocess
@@ -9,7 +13,7 @@ from .base import Channel
 class WeChatChannel(Channel):
     name = "wechat"
     description = "微信公众号文章"
-    backends = ["wechat-article-for-ai (Camoufox)"]
+    backends = ["wechat-article-for-ai (Camoufox)", "miku_ai (搜狗搜索)"]
     tier = 2
 
     def can_handle(self, url: str) -> bool:
@@ -18,15 +22,36 @@ 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_search = False
+
         try:
             import camoufox  # noqa: F401
+            has_read = True
         except ImportError:
+            pass
+
+        try:
+            import miku_ai  # noqa: F401
+            has_search = True
+        except ImportError:
+            pass
+
+        if has_read and has_search:
+            return "ok", "完整可用（搜索 + 阅读公众号文章）"
+        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"
+                "  pip install camoufox[geoip] markdownify beautifulsoup4 httpx mcp"
+            )
+        else:
             return "off", (
-                "需要 wechat-article-for-ai。安装：\n"
+                "需要安装微信公众号工具：\n"
+                "  # 阅读（URL → Markdown）：\n"
                 "  pip install camoufox[geoip] markdownify beautifulsoup4 httpx mcp\n"
-                "  # 或完整安装：\n"
-                "  git clone https://github.com/bzd6661/wechat-article-for-ai.git\n"
-                "  cd wechat-article-for-ai && pip install -r requirements.txt\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"
             )
-        return "ok", "可读取微信公众号文章（URL → Markdown）"
diff --git a/agent_reach/skill/SKILL.md b/agent_reach/skill/SKILL.md
index dc11870..b7039fa 100644
--- a/agent_reach/skill/SKILL.md
+++ b/agent_reach/skill/SKILL.md
@@ -253,21 +253,38 @@ mcporter call 'bosszhipin.get_job_detail_tool(job_url: "https://www.zhipin.com/j
 
 Fallback: `curl -s "https://r.jina.ai/https://www.zhipin.com/job_detail/xxx"`
 
-### 微信公众号 (wechat-article-for-ai)
+### 微信公众号 (wechat-article-for-ai + miku_ai)
 
-Uses Camoufox (stealth Firefox) to bypass WeChat's anti-bot detection and extract full article content.
+**Search** (miku_ai — Sogou WeChat search):
+
+```python
+# Search WeChat articles by keyword
+python3 -c "
+import asyncio
+from miku_ai import get_wexin_article
+
+async def search():
+    articles = await get_wexin_article('AI Agent', 5)
+    for a in articles:
+        print(f'{a[\"title\"]} | {a[\"source\"]} | {a[\"date\"]}')
+        print(f'  {a[\"url\"]}')
+
+asyncio.run(search())
+"
+```
+
+**Read** (Camoufox — stealth Firefox, bypasses WeChat anti-bot):
 
 ```bash
 # Read a WeChat article (returns Markdown with images)
 cd /path/to/wechat-article-for-ai && python3 main.py "https://mp.weixin.qq.com/s/ARTICLE_ID"
 
-# Batch convert multiple articles
-python3 main.py "URL1" "URL2" "URL3" -o ./output
-
 # Run as MCP server (for AI agent integration)
 python3 mcp_server.py
 ```
 
+Typical agent workflow: search → get URLs → immediately read full content.
+
 Note: WeChat articles require a real browser to render. Jina Reader and curl cannot read them.
 
 ### RSS (feedparser)