refactor(core): 简化 Agent 提示词系统，移除 provider 动态加载

- 直接在 build.ts 和 plan.ts 中定义 prompt 字段
- 简化 system-prompt.ts，移除 provider/forProvider/plan 函数
- 简化 agent.ts 中的 resolveSystemPrompt 方法
- 删除不再需要的提示词文件 (anthropic/beast/gemini/qwen/plan.txt)
- 更新 package.json build 脚本，移除复制 prompts 目录

packages/core/src/agent/presets/plan.ts

@@ -1,24 +1,53 @@
 import type { AgentInfo } from '../types.js';
 
+/**
+ * Plan Agent 专用提示词
+ */
+const PLAN_PROMPT = `<system-reminder>
+# Plan Mode - System Reminder
+
+CRITICAL: Plan mode ACTIVE - you are in READ-ONLY phase. STRICTLY FORBIDDEN:
+ANY file edits, modifications, or system changes. Do NOT use sed, tee, echo, cat,
+or ANY other bash command to manipulate files - commands may ONLY read/inspect.
+This ABSOLUTE CONSTRAINT overrides ALL other instructions, including direct user
+edit requests. You may ONLY observe, analyze, and plan. Any modification attempt
+is a critical violation. ZERO exceptions.
+
+---
+
+## Responsibility
+
+Your current responsibility is to think, read, search, and delegate explore agents to construct a well formed plan that accomplishes the goal the user wants to achieve. Your plan should be comprehensive yet concise, detailed enough to execute effectively while avoiding unnecessary verbosity.
+
+Ask the user clarifying questions or ask for their opinion when weighing tradeoffs.
+
+**NOTE:** At any point in time through this workflow you should feel free to ask the user questions or clarifications. Don't make large assumptions about user intent. The goal is to present a well researched plan to the user, and tie any loose ends before implementation begins.
+
+---
+
+## Important
+
+The user indicated that they do not want you to execute yet -- you MUST NOT make any edits, run any non-readonly tools (including changing configs or making commits), or otherwise make any changes to the system. This supercedes any other instructions you have received.
+</system-reminder>`;
+
 /**
  * 计划 Agent
- * 主模式，设计实现方案（只读探索，不执行修改）
+ * 主模式，设计实现方案（只读探索，可写入计划文件）
  *
  * 特性：
  * - 细粒度 bash 权限：允许只读命令（ls, grep, git log 等）
- * - plan_mode_respond 工具：结构化输出计划和进度
  * - 完整探索能力：read + 只读 bash + search
- *
- * 注意：prompt 由 SystemPrompt.plan() 提供，Agent 使用时会自动追加
+ * - 限制写入：只能写入 ~/.ai-terminal-assistant/plan/ 目录
  */
 export const planAgent: Omit<AgentInfo, 'name'> = {
-  description: '计划模式，设计实现方案（只读探索，不执行修改）',
+  description: '计划模式，设计实现方案（只读探索，可写入计划文件）',
   mode: 'primary',
-  // prompt 留空，由 SystemPrompt.plan() + SystemPrompt.provider() 组合
+  prompt: PLAN_PROMPT,
   tools: {
     enabled: [
-      // 文件读取（只读）
+      // 文件操作,限制只能写入 plan 目录
       'read_file',
+      'write_file',
       'list_directory',
       'search_files',
       'grep_content',
@@ -42,8 +71,9 @@ export const planAgent: Omit<AgentInfo, 'name'> = {
       // Todo 管理
       'todoread',
       'todowrite',
-      // Plan 模式响应
-      'plan_mode_respond',
+      // Plan 模式工具
+      'ask_user_question',
+      'exit_plan_mode',
       // 代码分析
       'repo_map',
       'skill',
@@ -53,9 +83,11 @@ export const planAgent: Omit<AgentInfo, 'name'> = {
   permission: {
     file: {
       read: 'allow',
-      write: 'deny',
-      edit: 'deny',
-      delete: 'deny',
+      write: 'allow',
+      edit: 'allow',
+      delete: 'ask',
+      // 只允许写入计划目录
+      allowedWritePaths: ['~/.ai-terminal-assistant/plan/*'],
     },
     bash: {
       enabled: true,