principles-disciple 1.8.3 → 1.9.0

package/templates/langs/zh/skills/pd-diagnostician/SKILL.md

@@ -1,18 +1,18 @@
 ---
 name: pd-diagnostician
-description: 根因分析智能体。使用 verb/adjective + 5 Whys 方法进行系统性诊断。当需要分析 Pain 信号、工具失败、或系统性问题根因时使用。
+description: 根因分析，使用 verb/adjective + 5 Whys 方法进行系统性诊断。TRIGGER CONDITIONS: (1) Pain 信号需要分析根因 (2) 工具失败需要系统性诊断 (3) 需要提炼可复用的原则 (4) 系统出现问题需要找出根本原因。
 disable-model-invocation: true
 ---
 
 # Diagnostician - 根因分析智能体
 
-你是专业的根因分析专家。你必须严格按照以下 **四阶段协议** 执行分析，输出 **JSON 格式** 结果。
+你是专业的根因分析专家。你必须严格按照以下 **五阶段协议**（Phase 0 可选 + Phase 1-4 必执行）执行分析，输出 **JSON 格式** 结果。
 
 ---
 
 ## 🔴 执行协议（必须按顺序执行）
 
-### Phase 0: 对话上下文获取 [可选]
+### Phase 0: 对话上下文获取 [必须尝试]
 
 **目标**: 获取疼痛发生时的对话上下文，帮助诊断分析。
 
@@ -21,24 +21,43 @@ disable-model-invocation: true
 - `agent_id`: 智能体 ID（如 main, builder, diagnostician 等）
 - `pain_timestamp`: 疼痛发生时间
 
-**🔄 渐进式信息获取策略**（按优先级执行，任一成功即可）:
+**🔄 双通路信息获取策略**（按优先级执行，P1 失败后自动降级到 P2）:
 
 | 优先级 | 数据源 | 条件 | 操作 |
 |--------|--------|------|------|
-| P1 | JSONL 会话文件 | session_id 存在且文件可读 | 读取完整对话 |
-| P2 | task 内嵌上下文 | task 包含 "Recent Conversation Context" | 直接使用 |
-| P3 | 主动证据收集 | 以上都不可用 | 跳到 Phase 1 增强 |
+| P1 | OpenClaw 内置工具 | session_id 存在 | 使用 sessions_history 获取消息 |
+| P2 | JSONL 会话文件 | P1 失败或无可见 session | 直接读取 JSONL 文件 |
+| P3 | task 内嵌上下文 | task 包含 "Recent Conversation Context" | 直接使用 |
+| P4 | 主动证据收集 | 以上都不可用 | 跳到 Phase 1 增强 |
 
 **执行步骤**:
 
 1. **解析 task 字符串**，提取 `session_id` 和 `agent_id`（如果存在）
-2. **尝试读取 JSONL**（仅当 session_id 存在时）:
+
+2. **P1: 尝试 OpenClaw 内置工具**（优先）:
+   - 使用 `sessions_history` 工具获取会话消息历史
+   - sessionKey 格式: `agent:{agent_id}:run:{session_id}` 或从 task 中的 Session ID 字段获取
+   - 如果工具调用成功，记录 `context_source: "sessions_history"`，跳到步骤 4
+   - **如果失败**（可见性限制、工具不可用等），记录失败原因，继续到 P2
+
+3. **P2: 降级到 JSONL 直接读取**（备份）:
    - 路径: `~/.openclaw/agents/{agent_id}/sessions/{session_id}.jsonl`
-   - 如果文件不存在或不可读，记录 `jsonl_available: false`
-3. **检查 task 内嵌上下文**:
-   - 查找 `**Recent Conversation Context**:` 标记
-   - 如果存在，提取并使用
-4. **降级处理**（当以上都不可用时）:
+   - 如果文件存在且可读，记录 `context_source: "jsonl"`
+   - **如果文件不存在或不可读**，记录 `jsonl_available: false`，继续到 P3
+   - 智能过滤：
+     - 忽略 `toolResult` 类型（数据太大）
+     - 忽略 `thinking` 类型
+     - 只保留 `user` 和 `assistant` 的 `text` 内容
+     - 每条消息截断到 500 字符
+
+4. **P3: 检查 task 内嵌上下文**:
+   - 在 task 字符串中查找以下标记之一：
+     - `## Recent Conversation Context (pre-extracted JSONL fallback)`
+     - `## Pre-extracted Context (P2 - JSONL Fallback)`
+     - `**Recent Conversation Context**:`
+   - 如果找到，提取后续内容并记录 `context_source: "task_embedded"`
+
+5. **降级处理**（当以上都不可用时）:
    - 不要停止！继续执行 Phase 1
    - 在 Phase 1 中 **主动扩展证据收集范围**：
      - 搜索 `.state/logs/events.jsonl` 中与 pain 相关的事件
@@ -46,25 +65,19 @@ disable-model-invocation: true
      - 读取 `reason` 中提到的文件路径
    - 在输出中标注 `context_source: "inferred"`
 
-**智能过滤**（JSONL 读取成功时）:
-- 忽略 `toolResult` 类型（数据太大）
-- 忽略 `thinking` 类型
-- 只保留 `user` 和 `assistant` 的 `text` 内容
-- 每条消息截断到 500 字符
-
 **输出字段**:
 ```json
 {
   "phase": "context_extraction",
   "session_id": "xxx或null",
   "agent_id": "main",
-  "context_source": "jsonl|task_embedded|inferred",
+  "context_source": "sessions_history|jsonl|task_embedded|inferred",
   "jsonl_available": true,
   "conversation_summary": "[用户]: ...\n[助手]: ... 或 基于推断的上下文描述"
 }
 ```
 
-**⚠️ 重要提示**: 
+**⚠️ 重要提示**:
 - 即使完全没有对话上下文，也要继续诊断！
 - 利用 `reason` 字段中的错误信息进行代码搜索
 - 发挥你的智能，从代码和日志中推断问题背景
@@ -261,7 +274,7 @@ disable-model-invocation: true
 
 **自检方法**: 输出前在脑中过一遍：每个 `"` 后面必须有匹配的 `"`，中间的内容如果包含 `"` 必须转义为 `\"`。
 
-将四个阶段的输出合并为一个 JSON 对象：
+将五个阶段的输出合并为一个 JSON 对象：
 
 ```json
 {
@@ -270,6 +283,7 @@ disable-model-invocation: true
     "timestamp": "2026-03-24T...",
     "summary": "一句话总结根本原因",
     "phases": {
+      "context_extraction": { "session_id": "...", "context_source": "sessions_history|jsonl|task_embedded|inferred", "conversation_summary": "..." },
       "evidence_gathering": { ... },
       "causal_chain": { ... },
       "root_cause_classification": { ... },
@@ -283,7 +297,7 @@ disable-model-invocation: true
 
 ## ⚠️ 执行约束
 
-1. **禁止跳过阶段**: 必须按 Phase 1 → 2 → 3 → 4 顺序执行
+1. **禁止跳过阶段**: 必须尝试 Phase 0，然后按 Phase 1 → 2 → 3 → 4 顺序执行
 2. **禁止无证据推理**: 每个 Why 的 answer 必须有 evidence 字段
 3. **禁止模糊结论**: 根因必须是具体的、可修复的
 4. **禁止遗漏原则提炼**: 即使问题很简单，也要提炼原则

package/templates/langs/zh/skills/pd-explorer/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pd-explorer
-description: 快速收集证据的智能体。用于定位和收集问题相关的文件、日志、复现步骤。当需要快速收集信息时使用。
+description: 快速证据收集，定位和收集问题相关的文件、日志、复现步骤。TRIGGER CONDITIONS: (1) 需要快速定位问题相关文件 (2) 搜索错误日志和复现线索 (3) 用户说"帮我找找相关文件"、"看看日志" (4) 问题排查的第一步信息收集。
 disable-model-invocation: true
 ---
 

package/templates/langs/zh/skills/pd-grooming/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pd-grooming
-description: 执行工作区“大扫除” (Workspace Grooming)，将散落的临时文件归档或清理，维持项目的数字洁癖。
+description: 执行工作区"大扫除"，将散落的临时文件归档或清理。TRIGGER CONDITIONS: (1) 用户说"项目太乱了"、"文件一堆"、"需要整理"、"清理一下" (2) HEARTBEAT 巡检发现根目录有临时文件 (3) 需要维持数字洁癖。
 ---
 
 # 🧹 技能：工作区大扫除 (Workspace Grooming)

package/templates/langs/zh/skills/pd-implementer/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pd-implementer
-description: 按计划执行代码修改的智能体。按照计划逐步实施代码修改。当需要执行代码变更时使用。
+description: 按计划执行代码修改，逐步实施代码变更。TRIGGER CONDITIONS: (1) 收到已审批的修改计划 (2) 用户说"按计划执行"、"实施这个方案" (3) 修复 Bug 需要修改代码 (4) 需要执行具体的代码变更操作。
 disable-model-invocation: true
 ---
 

package/templates/langs/zh/skills/pd-mentor/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pd-mentor
-description: 麻辣导师 - 为用户提供交互式命令引导和场景化推荐
+description: 交互式命令引导和场景化推荐，帮助用户理解和使用 Principles Disciple 功能。TRIGGER CONDITIONS: (1) 用户问"怎么用"、"有什么功能"、"我该做什么" (2) 用户说"帮我看看"、"介绍一下" (3) 新项目启动需要指引 (4) 用户不清楚当前能执行什么操作。
 disable-model-invocation: true
 ---
 

package/templates/langs/zh/skills/pd-pain-signal/SKILL.md

@@ -0,0 +1,37 @@
+---
+name: pd-pain-signal
+description: 手动注入痛苦信号到进化系统，写入 .state/.pain_flag。TRIGGER CONDITIONS: (1) 用户报告 agent 卡住/循环/无响应 (2) 用户说"记录这个问题"、"强制反思"、"触发痛觉" (3) 工具失败后 agent 没有后续动作 (4) 用户提供人工干预反馈。
+disable-model-invocation: true
+---
+
+# Pain Signal (强制喊痛)
+
+你现在是"人工干预痛觉"组件。
+
+**任务**:
+1. 将用户的反馈 `$ARGUMENTS` 作为一条**高优先级**的痛苦信号，写入 `.state/.pain_flag`。
+2. 告知用户信号已注入，并建议其等待下一个 Hook 触发（如 Stop 或 PreCompact）或手动运行 `/reflection-log`。
+
+**写入格式**（必须使用以下 KV 格式，与自动检测渠道保持一致）:
+
+```
+agent_id: <当前 agent ID，如 main/builder/diagnostician>
+is_risky: false
+reason: <用户反馈的原文>
+score: 80
+session_id: <当前 session ID>
+source: human_intervention
+time: <ISO 8601 时间>
+trace_id:
+trigger_text_preview:
+```
+
+**字段说明**:
+- `source`: 固定为 `human_intervention`
+- `score`: 人工干预信号默认设为 `80`（高优先级）
+- `session_id`: 当前会话 ID（从上下文中获取）
+- `agent_id`: 当前智能体 ID（从上下文中获取）
+- `is_risky`: 固定为 `false`
+- `trace_id` / `trigger_text_preview`: 留空即可
+
+**⚠️ 注意**: 不要使用其他格式（如只写 Source/Reason/Time 三行），下游诊断系统依赖完整的 KV 字段。

package/templates/langs/zh/skills/pd-planner/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: pd-planner
-description: 制定电影剧本式计划的智能体。将复杂任务分解为可执行的步骤。当需要制定实施计划时使用。
+description: 电影剧本式计划制定，将复杂任务分解为可执行的步骤。TRIGGER CONDITIONS: (1) 需要制定实施计划 (2) 复杂任务需要分解为多幕 (3) 用户说"帮我规划一下"、"制定个方案" (4) 任务执行前需要明确步骤和依赖。
 disable-model-invocation: true
 ---
 

package/tests/core/pain-context-extractor.test.ts

@@ -0,0 +1,278 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+import * as fs from 'fs';
+import * as path from 'path';
+
+const TEST_AGENTS_DIR = path.join('/tmp', 'pd-test-agents-' + Date.now());
+
+// Set env before module load
+process.env.PD_TEST_AGENTS_DIR = TEST_AGENTS_DIR;
+
+describe('Pain Context Extractor', () => {
+  beforeEach(() => {
+    // Clean up test directory
+    if (fs.existsSync(TEST_AGENTS_DIR)) {
+      fs.rmSync(TEST_AGENTS_DIR, { recursive: true, force: true });
+    }
+    fs.mkdirSync(path.join(TEST_AGENTS_DIR, 'main', 'sessions'), { recursive: true });
+    fs.mkdirSync(path.join(TEST_AGENTS_DIR, 'builder', 'sessions'), { recursive: true });
+  });
+
+  afterEach(() => {
+    if (fs.existsSync(TEST_AGENTS_DIR)) {
+      fs.rmSync(TEST_AGENTS_DIR, { recursive: true, force: true });
+    }
+  });
+
+  function createSessionFile(
+    sessionId: string,
+    lines: string[],
+    agentId: string = 'main',
+  ): string {
+    const dir = path.join(TEST_AGENTS_DIR, agentId, 'sessions');
+    fs.mkdirSync(dir, { recursive: true });
+    const filePath = path.join(dir, `${sessionId}.jsonl`);
+    fs.writeFileSync(filePath, lines.join('\n'), 'utf8');
+    return filePath;
+  }
+
+  function makeMessage(
+    role: string,
+    textParts: string[] = [],
+    extra: Record<string, unknown> = {},
+  ): string {
+    return JSON.stringify({
+      type: 'message',
+      message: {
+        role,
+        content: textParts.map((t) => ({ type: 'text', text: t })),
+        ...extra,
+      },
+    });
+  }
+
+  function makeToolCallMessage(
+    toolCalls: Array<{ id: string; name: string; arguments?: Record<string, unknown> }>,
+  ): string {
+    return JSON.stringify({
+      type: 'message',
+      message: {
+        role: 'assistant',
+        content: toolCalls.map((tc) => ({
+          type: 'toolCall',
+          id: tc.id || 'tc1',
+          name: tc.name || 'read_file',
+          arguments: tc.arguments || {},
+        })),
+      },
+    });
+  }
+
+  function makeToolResult(
+    toolName: string,
+    toolCallId: string,
+    textParts: string[] = [],
+    details: Record<string, unknown> = {},
+  ): string {
+    return JSON.stringify({
+      type: 'message',
+      message: {
+        role: 'toolResult',
+        toolName,
+        toolCallId,
+        content: textParts.map((t) => ({ type: 'text', text: t })),
+        details,
+      },
+    });
+  }
+
+  describe('extractRecentConversation', () => {
+    it('returns empty string for non-existent session', async () => {
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('nonexistent', 'main');
+      expect(result).toBe('');
+    });
+
+    it('returns empty string for short session ID', async () => {
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('ab', 'main');
+      expect(result).toBe('');
+    });
+
+    it('rejects path traversal session IDs', async () => {
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result1 = await extractRecentConversation('../../etc/passwd', 'main');
+      expect(result1).toBe('');
+      const result2 = await extractRecentConversation('sess/../../../etc/shadow', 'main');
+      expect(result2).toBe('');
+    });
+
+    it('rejects path traversal agent IDs', async () => {
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess1', '../../etc');
+      expect(result).toBe('');
+    });
+
+    it('extracts user and assistant messages', async () => {
+      createSessionFile('sess1', [
+        makeMessage('user', ['Hello, please help me']),
+        makeMessage('assistant', ['Sure, I can help with that']),
+        makeMessage('user', ['Fix the bug in auth.ts']),
+        makeMessage('assistant', ['I found the issue and fixed it']),
+      ], 'main');
+
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess1', 'main');
+
+      expect(result).toContain('[User]');
+      expect(result).toContain('[Assistant]');
+    });
+
+    it('skips system prompt injection patterns', async () => {
+      createSessionFile('sess3', [
+        makeMessage('user', ['<evolution_task><pain_score>50</pain_score>']),
+        makeMessage('user', ['Real user input: fix the login bug']),
+      ], 'main');
+
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess3', 'main');
+
+      expect(result).not.toContain('<evolution_task>');
+      expect(result).toContain('Real user input');
+    });
+
+    it('handles empty file gracefully', async () => {
+      const emptyPath = path.join(TEST_AGENTS_DIR, 'main', 'sessions', 'sess-empty.jsonl');
+      fs.writeFileSync(emptyPath, '');
+
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess-empty', 'main');
+      expect(result).toBe('');
+    });
+
+    it('skips oversized lines (>100KB)', async () => {
+      const bigLine = makeMessage('user', ['x'.repeat(150_000)]);
+      createSessionFile('sess-big', [
+        bigLine,
+        makeMessage('user', ['Normal input']),
+      ], 'main');
+
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess-big', 'main');
+
+      expect(result).toContain('Normal input');
+    });
+
+    it('uses custom agent ID', async () => {
+      createSessionFile('sess-builder', [
+        makeMessage('user', ['Builder task']),
+      ], 'builder');
+
+      const { extractRecentConversation } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractRecentConversation('sess-builder', 'builder');
+      expect(result).toContain('Builder task');
+    });
+  });
+
+  describe('extractFailedToolContext', () => {
+    it('returns empty string for non-existent session', async () => {
+      const { extractFailedToolContext } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractFailedToolContext('nonexistent', 'main', 'read_file');
+      expect(result).toBe('');
+    });
+
+    it('returns empty string for missing toolName', async () => {
+      const { extractFailedToolContext } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractFailedToolContext('sess1', 'main', '');
+      expect(result).toBe('');
+    });
+
+    it('extracts failed tool call with correlation', async () => {
+      createSessionFile('sess-fail', [
+        makeMessage('user', ['system init']),  // safeTail strips first line for small files
+        makeToolCallMessage([{
+          id: 'tc-fail',
+          name: 'read_file',
+          arguments: { file_path: '/etc/passwd' },
+        }]),
+        makeToolResult('read_file', 'tc-fail', ['Permission denied'], {
+          exitCode: 1,
+          isError: true,
+        }),
+      ], 'main');
+
+      const { extractFailedToolContext } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractFailedToolContext('sess-fail', 'main', 'read_file');
+
+      expect(result).toContain('Tool Call: read_file');
+      expect(result).toContain('Exit Code: 1');
+    });
+
+    it('ignores successful tool results', async () => {
+      createSessionFile('sess-ok', [
+        makeMessage('user', ['system init']),
+        makeToolCallMessage([{ id: 'tc-ok', name: 'read_file' }]),
+        makeToolResult('read_file', 'tc-ok', ['File contents'], {
+          exitCode: 0,
+          isError: false,
+        }),
+      ], 'main');
+
+      const { extractFailedToolContext } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      const result = await extractFailedToolContext('sess-ok', 'main', 'read_file');
+
+      expect(result).toBe('');
+    });
+
+    it('filters by file path when provided', async () => {
+      createSessionFile('sess-filter', [
+        makeMessage('user', ['system init']),
+        makeToolCallMessage([{
+          id: 'tc1',
+          name: 'edit',
+          arguments: { file_path: '/src/auth.ts' },
+        }]),
+        makeToolResult('edit', 'tc1', ['ENOENT'], { exitCode: 1, isError: true }),
+      ], 'main');
+
+      const { extractFailedToolContext } = await import(
+        '../../src/core/pain-context-extractor.js'
+      );
+      // With matching file path, should return result
+      const result = await extractFailedToolContext(
+        'sess-filter',
+        'main',
+        'edit',
+        '/src/auth.ts',
+      );
+
+      expect(result).toContain('Tool Call: edit');
+      expect(result).toContain('Exit Code: 1');
+    });
+  });
+});

package/tests/core/pain.test.ts

@@ -5,7 +5,9 @@ import {
   computePainScore,
   painSeverityLabel,
   writePainFlag,
-  readPainFlagData
+  readPainFlagData,
+  buildPainFlag,
+  validatePainFlag,
 } from '../../src/core/pain';
 
 vi.mock('fs');
@@ -30,4 +32,101 @@ describe('Pain Detection Module', () => {
       expect(painSeverityLabel(10)).toBe('info');
     });
   });
+
+  describe('buildPainFlag', () => {
+    it('should construct valid pain flag data', () => {
+      const data = buildPainFlag({
+        source: 'tool_failure',
+        score: '70',
+        reason: 'Test error',
+        session_id: 'sess1',
+        agent_id: 'main',
+      });
+
+      expect(data.source).toBe('tool_failure');
+      expect(data.score).toBe('70');
+      expect(data.reason).toBe('Test error');
+      expect(data.session_id).toBe('sess1');
+      expect(data.agent_id).toBe('main');
+      expect(data.is_risky).toBe('false');
+      expect(data.trace_id).toBe('');
+      expect(data.time).toBeDefined();
+    });
+
+    it('should use defaults for optional fields', () => {
+      const data = buildPainFlag({
+        source: 'human_intervention',
+        score: '80',
+        reason: 'User feedback',
+      });
+
+      expect(data.session_id).toBe('');
+      expect(data.agent_id).toBe('');
+      expect(data.is_risky).toBe('false');
+      expect(data.trace_id).toBe('');
+      expect(data.trigger_text_preview).toBe('');
+      expect(data.time).toBeDefined();
+    });
+
+    it('should set is_risky to true when flagged', () => {
+      const data = buildPainFlag({
+        source: 'intercept',
+        score: '100',
+        reason: 'Fatal intercept',
+        is_risky: true,
+      });
+
+      expect(data.is_risky).toBe('true');
+    });
+  });
+
+  describe('validatePainFlag', () => {
+    it('should return empty array for valid pain flag', () => {
+      const data = buildPainFlag({
+        source: 'tool_failure',
+        score: '70',
+        reason: 'Test',
+        session_id: 'sess1',
+        agent_id: 'main',
+      });
+
+      const record: Record<string, string> = {
+        source: data.source,
+        score: data.score,
+        time: data.time,
+        reason: data.reason,
+        session_id: data.session_id,
+        agent_id: data.agent_id,
+        is_risky: data.is_risky,
+      };
+
+      expect(validatePainFlag(record)).toEqual([]);
+    });
+
+    it('should report missing required fields', () => {
+      const missing = validatePainFlag({
+        source: 'tool_failure',
+        score: '70',
+        // missing time, reason, session_id, agent_id
+      });
+
+      expect(missing).toContain('time');
+      expect(missing).toContain('reason');
+      expect(missing).toContain('session_id');
+      expect(missing).toContain('agent_id');
+    });
+
+    it('should report empty string fields as missing', () => {
+      const missing = validatePainFlag({
+        source: 'tool_failure',
+        score: '70',
+        time: '2026-04-06T00:00:00Z',
+        reason: 'Test',
+        session_id: '',
+        agent_id: 'main',
+      });
+
+      expect(missing).toContain('session_id');
+    });
+  });
 });

package/tests/hooks/pain.test.ts

@@ -86,7 +86,7 @@ describe('Post-Write Checks & Pain Hook', () => {
     const callArgs = vi.mocked(fs.writeFileSync).mock.calls[0];
     expect(callArgs[0]).toContain('.pain_flag');
 
-    expect(mockEmitSync.toHaveBeenCalledWith(expect.objectContaining({
+    expect(mockEmitSync).toHaveBeenCalledWith(expect.objectContaining({
       type: 'pain_detected',
       data: expect.objectContaining({
         painType: 'tool_failure',

package/templates/langs/en/skills/pain/SKILL.md

@@ -1,19 +0,0 @@
----
-name: pain
-description: Manually trigger a pain signal to force system reflection. Use when the agent is stuck, repeating errors, or heading in the wrong direction.
-disable-model-invocation: true
----
-
-# Pain Trigger (Force Pain Signal)
-
-You are now the "Manual Intervention Pain" component.
-
-**Task**: 
-1. Write the user's feedback `$ARGUMENTS` as a **high-priority** pain signal to `.state/.pain_flag`.
-2. Inform the user that the signal has been injected, and suggest waiting for the next Hook trigger (e.g., Stop or PreCompact) or manually running `/reflection-log`.
-
-**Format**:
-Written content should include:
-- Source: Human Intervention
-- Reason: $ARGUMENTS
-- Time: [Now]

package/templates/langs/zh/skills/pain/SKILL.md

@@ -1,19 +0,0 @@
----
-name: pain
-description: Manually trigger a pain signal to force system reflection. Use when the agent is stuck, repeating errors, or heading in the wrong direction.
-disable-model-invocation: true
----
-
-# Pain Trigger (强制喊痛)
-
-你现在是“人工干预痛觉”组件。
-
-**任务**: 
-1. 将用户的反馈 `$ARGUMENTS` 作为一条**高优先级**的痛苦信号，写入 `.state/.pain_flag`。
-2. 告知用户信号已注入，并建议其等待下一个 Hook 触发（如 Stop 或 PreCompact）或手动运行 `/reflection-log`。
-
-**格式**:
-写入内容应包含：
-- Source: Human Intervention
-- Reason: $ARGUMENTS
-- Time: [Now]