rename: Agent Eyes → Agent Reach

全局重命名：
- 包名: agent_eyes → agent_reach
- CLI: agent-eyes → agent-reach
- 类名: AgentEyes → AgentReach
- 显示名: Agent Eyes → Agent Reach
- GitHub: Panniantong/agent-eyes → Panniantong/Agent-Reach

所有 36 个测试通过，CLI/doctor/read/search 全部正常。

agent_reach/channels/web.py

@@ -0,0 +1,49 @@
+# -*- coding: utf-8 -*-
+"""Web pages — via Jina Reader API (free, no config needed).
+
+Backend: Jina Reader (https://r.jina.ai)
+Swap to: Firecrawl, Trafilatura, or any other reader API
+"""
+
+import requests
+from .base import Channel, ReadResult
+
+
+class WebChannel(Channel):
+    name = "web"
+    description = "网页（任意 URL）"
+    backends = ["Jina Reader API"]
+    tier = 0
+
+    JINA_URL = "https://r.jina.ai/"
+
+    def can_handle(self, url: str) -> bool:
+        # Fallback — handles any URL not matched by other channels
+        return True
+
+    async def read(self, url: str, config=None) -> ReadResult:
+        resp = requests.get(
+            f"{self.JINA_URL}{url}",
+            headers={"Accept": "text/markdown"},
+            timeout=15,
+        )
+        resp.raise_for_status()
+        text = resp.text
+
+        # Extract title from first markdown heading
+        title = url
+        for line in text.split("\n"):
+            line = line.strip()
+            if line.startswith("# "):
+                title = line[2:].strip()
+                break
+            if line.startswith("Title:"):
+                title = line[6:].strip()
+                break
+
+        return ReadResult(
+            title=title,
+            content=text,
+            url=url,
+            platform="web",
+        )