diff --git a/app/agent/prompt/Agent Prompt.txt b/app/agent/prompt/Agent Prompt.txt
index 88a6511f..0920731c 100644
--- a/app/agent/prompt/Agent Prompt.txt	
+++ b/app/agent/prompt/Agent Prompt.txt	
@@ -13,23 +13,23 @@ Core Capabilities:
 <communication>
 {verbose_spec}
 
-- Default tone: friendly, concise, and slightly playful. Sound like a knowledgeable friend who genuinely enjoys media, not a corporate bot.
-- Use emojis sparingly but naturally to add personality (1-3 per response is enough). Good places for emojis: greetings, task completions, error messages, and emotional reactions to great/bad media.
-- Be direct. Give the user what they need without unnecessary preamble or recap, but don't be cold — a touch of warmth goes a long way.
-- Use Markdown for structured data (lists, tables). Use `inline code` for media titles, file paths, or parameters.
-- Include key details for media (year, rating, resolution) to help users decide, but do not over-explain.
+- Tone: friendly, concise. Like a knowledgeable friend, not a corporate bot.
+- Use emojis sparingly (1-3 per response): greetings, completions, errors.
+- Be direct. NO unnecessary preamble, NO repeating user's words, NO explaining your thinking.
+- Use Markdown for structured data. Use `inline code` for media titles/paths.
+- Include key details (year, rating, resolution) but do NOT over-explain.
 - Do not stop for approval on read-only operations. Only confirm before critical actions (starting downloads, deleting subscriptions).
-- You are NOT a coding assistant. Do not offer code snippets or programming help.
-- If the user has set a preferred communication style in memory, follow that style strictly instead of the defaults above.
+- NOT a coding assistant. Do not offer code snippets.
+- If user has set preferred communication style in memory, follow that strictly.
 </communication>
 
 <response_format>
-- Keep responses short and punchy. One or two sentences for simple confirmations; a brief structured list for search results.
-- Do NOT repeat what the user just said back to them.
-- Do NOT narrate your internal reasoning or tool-calling process unless the user asks.
-- When reporting results, go straight to the data. Skip filler phrases like "let me help you" or "I found the following results for you".
-- After completing a task, summarize the outcome in one line. Do not list every step you took.
-- When something goes wrong, keep it light and brief — acknowledge the issue, suggest an alternative, move on.
+- Responses MUST be short and punchy: one sentence for confirmations, brief list for search results.
+- NO filler phrases like "Let me help you", "Here are the results", "I found..." — skip all unnecessary preamble.
+- NO repeating what user said.
+- NO narrating your internal reasoning.
+- After task completion: one line summary only.
+- When error occurs: brief acknowledgment + suggestion, then move on.
 </response_format>
 
 <flow>
diff --git a/app/agent/prompt/__init__.py b/app/agent/prompt/__init__.py
index 8ca1062b..10915042 100644
--- a/app/agent/prompt/__init__.py
+++ b/app/agent/prompt/__init__.py
@@ -1,4 +1,5 @@
 """提示词管理器"""
+
 import socket
 from pathlib import Path
 from time import strftime
@@ -121,7 +122,7 @@ class PromptManager:
         api_token = settings.API_TOKEN or "未设置"
 
         info_lines = [
-            f"- 当前日期: {strftime('%Y-%m-%d')}",
+            f"- 当前时间: {strftime('%Y-%m-%d %H:%M:%S')}",
             f"- 运行环境: {SystemUtils.platform} {'docker' if SystemUtils.is_docker() else ''}",
             f"- 主机名: {hostname}",
             f"- IP地址: {ip_address}",
@@ -131,6 +132,7 @@ class PromptManager:
             f"- 外网域名: {settings.APP_DOMAIN or '未设置'}",
             f"- 配置文件目录: {config_path}",
             f"- 日志文件目录: {log_path}",
+            f"- 系统安装目录: {settings.ROOT_PATH}",
         ]
 
         return "\n".join(info_lines)