diff --git a/scripts/rsshub-zhiyi-fetch.py b/scripts/rsshub-zhiyi-fetch.py
index d969032c..1d046235 100644
--- a/scripts/rsshub-zhiyi-fetch.py
+++ b/scripts/rsshub-zhiyi-fetch.py
@@ -20,6 +20,7 @@ rsshub-zhiyi-fetch.py — RSSHub GitHub AI 内容抓取 → 写入织忆
 import json
 import os
 import sys
+import time
 import urllib.request
 import xml.etree.ElementTree as ET
 from datetime import datetime
@@ -40,11 +41,20 @@ SOURCES = [
     ("/github/search/AI+agents", "GitHub AI Agents"),
 ]
 
-def fetch_rss(path):
+def fetch_rss(path, retries=3):
+    """抓取 RSS，带重试。2026-08-05: github.com:443 瞬时抖动曾导致三个源 503，加重试容错。"""
     url = RSSHUB_BASE + path
-    req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
-    with urllib.request.urlopen(req, timeout=30) as resp:
-        return resp.read().decode("utf-8", "ignore")
+    last_exc = None
+    for attempt in range(retries):
+        try:
+            req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+            with urllib.request.urlopen(req, timeout=45) as resp:
+                return resp.read().decode("utf-8", "ignore")
+        except Exception as e:
+            last_exc = e
+            if attempt < retries - 1:
+                time.sleep(3 * (attempt + 1))
+    raise last_exc
 
 def parse_items(xml_text):
     """解析 RSS 2.0 / Atom，返回 [{title, link, desc}]"""
diff --git a/skills/.usage.json b/skills/.usage.json
index 48c44ffe..c45bc8e2 100644
--- a/skills/.usage.json
+++ b/skills/.usage.json
@@ -510,14 +510,14 @@
     "archived_at": null,
     "created_at": "2026-07-28T14:02:19.195060+00:00",
     "created_by": "agent",
-    "last_patched_at": "2026-08-03T14:02:56.602777+00:00",
-    "last_used_at": "2026-08-03T14:02:48.782592+00:00",
-    "last_viewed_at": "2026-08-03T14:02:48.733419+00:00",
-    "patch_count": 5,
+    "last_patched_at": "2026-08-05T14:02:32.841670+00:00",
+    "last_used_at": "2026-08-05T14:02:28.542800+00:00",
+    "last_viewed_at": "2026-08-05T14:02:28.539531+00:00",
+    "patch_count": 6,
     "pinned": false,
     "state": "active",
-    "use_count": 6,
-    "view_count": 6
+    "use_count": 7,
+    "view_count": 7
   },
   "design-feasibility-review": {
     "archived_at": null,
diff --git a/skills/productivity/daily-recap/SKILL.md b/skills/productivity/daily-recap/SKILL.md
index d1d7b3ca..813eb00c 100644
--- a/skills/productivity/daily-recap/SKILL.md
+++ b/skills/productivity/daily-recap/SKILL.md
@@ -37,6 +37,13 @@ tags: [复盘, 每日, 自动化]
 6. **发送** 通过飞书 `send_message`（或直接返回）将报告推送给用户。
 
 ## 常见陷阱
+- 会话记录为空时 → 使用占位文本 “暂无会话记录”。
+- 守护日志文件过大 → 只读取最近 100 行，以防 `read_file` 超限。
+- 心迹日志可能缺少 `type` 字段 → 默认视为 `moment`，安全提取 `content`。
+- 若任何来源为空，报告仍保持完整结构，使用占位文本。
+- **新增**：确保会话记录功能已启用，否则 `session_search` 返回空，报告会缺失关键信息。请在 Hermes 配置中打开 `session_logging`。
+- 若报告重复出现步骤说明，检查 SKILL.md 是否出现重复段落，必要时删除冗余。
+
 - 会话记录为空时 → 使用占位文本 “暂无会话记录”。
 - 守护日志文件过大 → 只读取最近 100 行，以防 `read_file` 超限。
 - 心迹日志可能缺少 `type` 字段 → 默认视为 `moment`，安全提取 `content`。