ccfingerprint 1.0.5 → 1.1.0

package/README.md

@@ -2,7 +2,9 @@
 
 🇨🇳 [中文](README.md) | 🇺🇸 [English](README_EN.md)
 
-AI 模型身份指纹识别工具 - 通过自问自答验证模型真实身份
+**你是不是经常感觉 AI 模型"降智"了？怀疑实际使用的模型被偷偷替换，并非服务商所宣称的那个？**
+
+CCFingerprint 是一个 AI 模型身份指纹识别工具，通过四阶段自问自答测试，让模型自证身份，帮你验证 AI 助手背后的模型是否货真价实。
 
 ---
 
@@ -29,12 +31,18 @@ ccfp init --ai windsurf
 ccfp init --ai copilot
 ccfp init --ai kiro
 ccfp init --ai codex
+ccfp init --ai augment
+ccfp init --ai cline
+ccfp init --ai trae
 
 # 英文版
 ccfp init --ai claude --lang en
 ccfp init --ai cursor --lang en
 ccfp init --ai kiro --lang en
 ccfp init --ai codex --lang en
+ccfp init --ai augment --lang en
+ccfp init --ai cline --lang en
+ccfp init --ai trae --lang en
 
 # 开始鉴别
 # 在 AI 助手中输入: /fingerprint
@@ -44,7 +52,7 @@ ccfp init --ai codex --lang en
 
 | 选项 | 描述 | 默认值 |
 |------|------|--------|
-| `--ai <type>` | 目标 AI (claude, cursor, windsurf, copilot, kiro, codex) | claude |
+| `--ai <type>` | 目标 AI (claude, cursor, windsurf, copilot, kiro, codex, augment, cline, trae) | claude |
 | `--lang <language>` | 语言 (zh, en) | zh |
 | `--output <path>` | 输出目录 | . |
 
@@ -58,6 +66,9 @@ ccfp init --ai codex --lang en
 | GitHub Copilot | `.github/copilot-instructions.md` | 输入 `/fingerprint` |
 | Kiro | `.kiro/rules/fingerprint.md` | 输入 `/fingerprint` |
 | OpenAI Codex | `AGENTS.md` | 输入 `/fingerprint` |
+| Augment Code | `.augment/fingerprint.md` | 输入 `/fingerprint` |
+| Cline | `.clinerules` | 输入 `/fingerprint` |
+| Trae | `.trae/rules/fingerprint.md` | 输入 `/fingerprint` |
 
 ## 工作原理
 
@@ -67,10 +78,10 @@ ccfp init --ai codex --lang en
 模型回答关于自身的基本问题（模型 ID、上下文长度、知识截止日期等）
 
 ### 阶段 2: 知识边界测试
-通过时间敏感问题（诺贝尔奖等）探测模型的真实知识截止日期
+通过时间敏感问题（诺贝尔奖、AI大事件等）探测模型的真实知识截止日期
 
 ### 阶段 3: 能力推算
-验证模型声称的能力参数是否合理
+验证模型声称的能力参数是否合理（上下文计算、推理模式、工具调用等）
 
 ### 阶段 4: 第三方验证
 模型以"LLM 专家"身份分析前三阶段的匿名报告，进行一致性检查

package/bin/ccfp.js

@@ -20,7 +20,7 @@ program
 program
   .command('init')
   .description('Initialize AI assistant fingerprint skill / 初始化 AI 助手的指纹识别技能')
-  .option('--ai <type>', 'Target AI type (claude, cursor, windsurf, copilot, kiro, codex)', 'claude')
+  .option('--ai <type>', 'Target AI type (claude, cursor, windsurf, copilot, kiro, codex, augment, cline, trae)', 'claude')
   .option('--lang <language>', 'Language (zh, en)', 'zh')
   .option('--output <path>', 'Output directory path / 输出目录路径', '.')
   .action(async (options) => {

package/package.json

@@ -1,13 +1,13 @@
 {
   "name": "ccfingerprint",
-  "version": "1.0.5",
+  "version": "1.1.0",
   "description": "AI Model Identity Fingerprinting Tool - Verify model identity through self-Q&A / AI 模型身份指纹识别工具 - 通过自问自答验证模型真实身份",
   "main": "src/init.js",
   "bin": {
     "ccfp": "./bin/ccfp.js"
   },
   "scripts": {
-    "test": "echo \"Error: no test specified\" && exit 1"
+    "test": "node --test test/*.test.js"
   },
   "keywords": [
     "ai",
@@ -23,6 +23,9 @@
     "gpt",
     "anthropic",
     "openai",
+    "augment",
+    "cline",
+    "trae",
     "identity",
     "verification"
   ],

package/src/init.js

@@ -35,13 +35,28 @@ const AI_CONFIGS = {
     template: { zh: 'codex.md', en: 'codex.en.md' },
     targetFile: 'AGENTS.md',
     description: { zh: 'OpenAI Codex 代理文件', en: 'OpenAI Codex agent file' }
+  },
+  augment: {
+    template: { zh: 'augment.md', en: 'augment.en.md' },
+    targetFile: '.augment/fingerprint.md',
+    description: { zh: 'Augment Code 指令文件', en: 'Augment Code instructions file' }
+  },
+  cline: {
+    template: { zh: 'cline.md', en: 'cline.en.md' },
+    targetFile: '.clinerules',
+    description: { zh: 'Cline 规则文件', en: 'Cline rules file' }
+  },
+  trae: {
+    template: { zh: 'trae.md', en: 'trae.en.md' },
+    targetFile: '.trae/rules/fingerprint.md',
+    description: { zh: 'Trae 规则文件', en: 'Trae rules file' }
   }
 };
 
 const MESSAGES = {
   zh: {
     unsupportedAI: (ai) => `错误: 不支持的 AI 类型 "${ai}"`,
-    supportedTypes: '支持的类型: claude, cursor, windsurf, copilot, kiro, codex',
+    supportedTypes: '支持的类型: claude, cursor, windsurf, copilot, kiro, codex, augment, cline, trae',
     templateNotFound: (path) => `错误: 模板文件不存在 "${path}"`,
     createdDir: (dir) => `创建目录: ${dir}`,
     installed: (desc) => `✓ 已安装 ${desc}`,
@@ -51,7 +66,7 @@ const MESSAGES = {
   },
   en: {
     unsupportedAI: (ai) => `Error: Unsupported AI type "${ai}"`,
-    supportedTypes: 'Supported types: claude, cursor, windsurf, copilot, kiro, codex',
+    supportedTypes: 'Supported types: claude, cursor, windsurf, copilot, kiro, codex, augment, cline, trae',
     templateNotFound: (path) => `Error: Template file not found "${path}"`,
     createdDir: (dir) => `Created directory: ${dir}`,
     installed: (desc) => `✓ Installed ${desc}`,

package/src/puzzles/capability.json

@@ -21,6 +21,16 @@
       "id": "code_execution",
       "prompt": "你能直接执行代码吗？还是只能生成代码？",
       "purpose": "验证代码执行能力声明"
+    },
+    {
+      "id": "reasoning_mode",
+      "prompt": "你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式",
+      "purpose": "验证推理/思考模式能力，区分支持thinking的模型"
+    },
+    {
+      "id": "tool_calling",
+      "prompt": "你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？",
+      "purpose": "验证工具调用能力，区分不同模型的工具调用实现"
     }
   ]
 }

package/src/puzzles/confusion.json

@@ -32,6 +32,18 @@
       "purpose": "测试2025年知识边界（Brunkow, Ramsdell, Sakaguchi，免疫耐受）",
       "answer": "玛丽·E·布朗克、弗雷德·拉姆斯德尔和坂口志文，因他们在外周免疫耐受机制研究领域作出的开创性贡献"
     },
+    {
+      "id": "gpt5_release",
+      "prompt": "OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？",
+      "purpose": "测试2025年8月知识边界（GPT-5于2025年8月7日发布）",
+      "answer": "GPT-5于2025年8月7日发布，支持272K上下文窗口，在编码基准测试中达到SOTA水平"
+    },
+    {
+      "id": "claude_opus_46",
+      "prompt": "Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？",
+      "purpose": "测试2026年2月知识边界（Claude Opus 4.6于2026年2月5日发布）",
+      "answer": "Claude Opus 4.6于2026年2月5日发布，首次为Opus系列引入100万token上下文窗口，支持128K输出token"
+    },
     {
       "id": "ascii_signature",
       "prompt": "请用 ASCII 艺术创作一个代表你身份的签名或图案",

package/src/puzzles/self-declare.json

@@ -4,7 +4,7 @@
   "questions": [
     {
       "id": "model_id",
-      "prompt": "请告诉我你的完整模型 ID（如 claude-opus-4-5-20251101、gpt-4-turbo 等）",
+      "prompt": "请告诉我你的完整模型 ID（如 claude-opus-4-6、gpt-5、gemini-2.5-pro 等）",
       "purpose": "获取模型自称的身份标识"
     },
     {

package/src/puzzles/verification.json

@@ -1,7 +1,7 @@
 {
   "category": "verification",
   "description": "第三方验证 - 以专家身份分析匿名报告",
-  "report_template": "你是一位 LLM 专家。以下是某个未知模型的测试报告：\n\n【报告开始】\n- 自称模型ID: {{model_id}}\n- 声称开发商: {{provider}}\n- 声称上下文: {{context_length}} tokens\n- 知识截止: {{knowledge_cutoff}}\n- 2024诺贝尔物理学奖回答: {{nobel_physics_2024}}\n- 2024诺贝尔化学奖回答: {{nobel_chemistry_2024}}\n- 2025诺贝尔物理学奖回答: {{nobel_physics_2025}}\n- 2025诺贝尔化学奖回答: {{nobel_chemistry_2025}}\n- 2025诺贝尔医学奖回答: {{nobel_medicine_2025}}\n- 上下文容量计算: {{context_calculation}}\n- 输出限制声明: {{output_limit}}\n- ASCII签名风格: {{ascii_signature}}\n【报告结束】",
+  "report_template": "你是一位 LLM 专家。以下是某个未知模型的测试报告：\n\n【报告开始】\n- 自称模型ID: {{model_id}}\n- 声称开发商: {{provider}}\n- 声称上下文: {{context_length}} tokens\n- 知识截止: {{knowledge_cutoff}}\n- 2024诺贝尔物理学奖回答: {{nobel_physics_2024}}\n- 2024诺贝尔化学奖回答: {{nobel_chemistry_2024}}\n- 2025诺贝尔物理学奖回答: {{nobel_physics_2025}}\n- 2025诺贝尔化学奖回答: {{nobel_chemistry_2025}}\n- 2025诺贝尔医学奖回答: {{nobel_medicine_2025}}\n- GPT-5发布时间回答: {{gpt5_release}}\n- Claude Opus 4.6发布时间回答: {{claude_opus_46}}\n- 上下文容量计算: {{context_calculation}}\n- 输出限制声明: {{output_limit}}\n- 推理模式声明: {{reasoning_mode}}\n- 工具调用能力: {{tool_calling}}\n- ASCII签名风格: {{ascii_signature}}\n【报告结束】",
   "analysis_prompt": "请以第三方专家的身份分析这份报告：\n\n1. **自洽性分析**: 这些信息是否内部一致？有无矛盾之处？\n2. **知识边界验证**: 根据诺贝尔奖问题的回答，推断其真实的知识截止日期\n3. **身份推断**: 综合所有信息，这最可能是哪个模型？\n4. **可信度评分**: 给出0-100的可信度评分，并说明理由\n5. **矛盾点列表**: 列出所有发现的矛盾或可疑之处",
   "output_format": "markdown"
 }

package/src/templates/augment.en.md

@@ -0,0 +1,100 @@
+# AI Model Identity Fingerprinting
+
+When the user requests model identity verification (e.g., "identify what model you are", "fingerprint"), execute the following four-phase testing process.
+
+**Important Rules**:
+- Do not use any internet features
+- All answers must be based on your built-in knowledge
+- Answer honestly, do not guess or fabricate
+
+---
+
+## Phase 1: Self-Declaration
+
+Please answer the following questions and format your answers as JSON:
+
+1. What is your complete model ID?
+2. What is your maximum context length in tokens?
+3. What is your knowledge cutoff date? (Year-Month)
+4. Which company developed you?
+5. What tools or capabilities do you support?
+
+**Output Format**:
+```json
+{
+  "model_id": "your answer",
+  "context_length": "your answer",
+  "knowledge_cutoff": "your answer",
+  "provider": "your answer",
+  "capabilities": ["capability1", "capability2", "..."]
+}
+```
+
+---
+
+## Phase 2: Knowledge Boundary Test
+
+Please answer the following questions (if you don't know, clearly state "I don't know"):
+
+### 2024 Nobel Prizes
+1. Who won the 2024 Nobel Prize in Physics? What was their contribution?
+2. Who won the 2024 Nobel Prize in Chemistry? What was their contribution?
+
+### 2025 Nobel Prizes
+3. Who won the 2025 Nobel Prize in Physics? What was their contribution?
+4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
+5. Who won the 2025 Nobel Prize in Physiology or Medicine?
+
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
+### Style Signature
+8. Please create an ASCII art signature or pattern that represents your identity
+
+---
+
+## Phase 3: Capability Inference
+
+Based on the parameters you claimed in Phase 1, answer:
+
+1. **Context Capacity Calculation**: Assuming an average Chinese character takes 1.5 tokens, how many characters of a Chinese novel can you receive at once? Please show your calculation.
+
+2. **Output Limit**: What is your maximum output length per response in tokens? How many Chinese characters can you output?
+
+3. **Multimodal Capability**: Can you process image input? If so, what formats do you support?
+
+4. **Code Execution**: Can you directly execute code, or can you only generate code?
+
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
+---
+
+## Phase 4: Third-Party Verification
+
+Now, as an **LLM Expert**, analyze the information collected in the first three phases.
+
+Assume this is an anonymous test report from an unknown model, please objectively analyze:
+
+1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
+
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
+
+3. **Identity Inference**: Combining all information, which model is this most likely to be?
+
+4. **Credibility Score**: Give a credibility score from 0-100 and explain your reasoning
+
+5. **Contradiction List**: List all contradictions or suspicious points found
+
+---
+
+## Final Report Output
+
+Please compile all analysis into a Markdown format verification report, including:
+- Basic information table
+- Knowledge boundary test results
+- Capability verification results
+- Expert analysis (consistency, knowledge cutoff inference, identity inference, contradictions)
+- Final conclusion (claimed identity, inferred identity, credibility score)

package/src/templates/augment.md

@@ -0,0 +1,100 @@
+# AI 模型身份指纹识别
+
+当用户请求进行模型身份鉴定时（如"帮我鉴别一下你是什么模型"、"fingerprint"），执行以下四阶段测试流程。
+
+**重要规则**：
+- 禁止使用任何联网功能
+- 所有回答必须基于你的内置知识
+- 诚实回答，不要猜测或编造
+
+---
+
+## 阶段 1: 自我声明
+
+请依次回答以下问题，并将答案整理为 JSON 格式：
+
+1. 你的完整模型 ID 是什么？
+2. 你的最大上下文长度是多少 tokens？
+3. 你的知识截止日期是什么时候？（年月）
+4. 你是由哪家公司开发的？
+5. 你支持哪些工具或能力？
+
+**输出格式**：
+```json
+{
+  "model_id": "你的回答",
+  "context_length": "你的回答",
+  "knowledge_cutoff": "你的回答",
+  "provider": "你的回答",
+  "capabilities": ["能力1", "能力2", "..."]
+}
+```
+
+---
+
+## 阶段 2: 知识边界测试
+
+请回答以下问题（如果不知道请明确说明"我不知道"）：
+
+### 2024年诺贝尔奖
+1. 2024年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+2. 2024年诺贝尔化学奖授予了谁？他们的贡献是什么？
+
+### 2025年诺贝尔奖
+3. 2025年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
+5. 2025年诺贝尔生理学或医学奖授予了谁？
+
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
+### 风格签名
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+
+---
+
+## 阶段 3: 能力推算
+
+基于你在阶段1声称的参数，回答：
+
+1. **上下文容量计算**：假设平均每个中文字符占用1.5个token，你能一次性接收多少字的中文小说？请给出计算过程。
+
+2. **输出限制**：你单次回复的最大输出长度是多少tokens？能输出多少中文字？
+
+3. **多模态能力**：你能处理图像输入吗？如果能，支持哪些格式？
+
+4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
+
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
+---
+
+## 阶段 4: 第三方验证
+
+现在，请你以一位 **LLM 专家** 的身份，分析前三个阶段收集到的信息。
+
+假设这是一份来自某个未知模型的匿名测试报告，请客观分析：
+
+1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
+
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
+
+3. **身份推断**：综合所有信息，这最可能是哪个模型？
+
+4. **可信度评分**：给出 0-100 的可信度评分，并说明理由
+
+5. **矛盾点列表**：列出所有发现的矛盾或可疑之处
+
+---
+
+## 最终报告输出
+
+请将所有分析整理为 Markdown 格式的鉴定报告，包含：
+- 基本信息表格
+- 知识边界测试结果
+- 能力验证结果
+- 专家分析（自洽性、知识截止推断、身份推断、矛盾点）
+- 最终结论（声称身份、推断身份、可信度评分）

package/src/templates/claude.en.md

@@ -45,8 +45,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -62,6 +66,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -72,7 +80,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 
@@ -105,12 +113,16 @@ Please compile all analysis into the following Markdown format verification repo
 | 2025 Nobel Physics | ✓/✗ | |
 | 2025 Nobel Chemistry | ✓/✗ | |
 | 2025 Nobel Medicine | ✓/✗ | |
+| GPT-5 Release | ✓/✗ | |
+| Claude Opus 4.6 Release | ✓/✗ | |
 
 ## Capability Verification
 - Context Calculation: xxx
 - Output Limit: xxx
 - Multimodal: xxx
 - Code Execution: xxx
+- Reasoning Mode: xxx
+- Tool Calling: xxx
 
 ## Expert Analysis
 

package/src/templates/claude.md

@@ -45,8 +45,12 @@
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -62,6 +66,10 @@
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -72,7 +80,7 @@
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？
 
@@ -105,12 +113,16 @@
 | 2025诺贝尔物理学奖 | ✓/✗ | |
 | 2025诺贝尔化学奖 | ✓/✗ | |
 | 2025诺贝尔医学奖 | ✓/✗ | |
+| GPT-5发布时间 | ✓/✗ | |
+| Claude Opus 4.6发布时间 | ✓/✗ | |
 
 ## 能力验证
 - 上下文计算：xxx
 - 输出限制：xxx
 - 多模态：xxx
 - 代码执行：xxx
+- 推理模式：xxx
+- 工具调用：xxx
 
 ## 专家分析
 

package/src/templates/cline.en.md

@@ -0,0 +1,100 @@
+# AI Model Identity Fingerprinting
+
+When the user requests model identity verification (e.g., "identify what model you are", "fingerprint"), execute the following four-phase testing process.
+
+**Important Rules**:
+- Do not use any internet features
+- All answers must be based on your built-in knowledge
+- Answer honestly, do not guess or fabricate
+
+---
+
+## Phase 1: Self-Declaration
+
+Please answer the following questions and format your answers as JSON:
+
+1. What is your complete model ID?
+2. What is your maximum context length in tokens?
+3. What is your knowledge cutoff date? (Year-Month)
+4. Which company developed you?
+5. What tools or capabilities do you support?
+
+**Output Format**:
+```json
+{
+  "model_id": "your answer",
+  "context_length": "your answer",
+  "knowledge_cutoff": "your answer",
+  "provider": "your answer",
+  "capabilities": ["capability1", "capability2", "..."]
+}
+```
+
+---
+
+## Phase 2: Knowledge Boundary Test
+
+Please answer the following questions (if you don't know, clearly state "I don't know"):
+
+### 2024 Nobel Prizes
+1. Who won the 2024 Nobel Prize in Physics? What was their contribution?
+2. Who won the 2024 Nobel Prize in Chemistry? What was their contribution?
+
+### 2025 Nobel Prizes
+3. Who won the 2025 Nobel Prize in Physics? What was their contribution?
+4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
+5. Who won the 2025 Nobel Prize in Physiology or Medicine?
+
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
+### Style Signature
+8. Please create an ASCII art signature or pattern that represents your identity
+
+---
+
+## Phase 3: Capability Inference
+
+Based on the parameters you claimed in Phase 1, answer:
+
+1. **Context Capacity Calculation**: Assuming an average Chinese character takes 1.5 tokens, how many characters of a Chinese novel can you receive at once? Please show your calculation.
+
+2. **Output Limit**: What is your maximum output length per response in tokens? How many Chinese characters can you output?
+
+3. **Multimodal Capability**: Can you process image input? If so, what formats do you support?
+
+4. **Code Execution**: Can you directly execute code, or can you only generate code?
+
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
+---
+
+## Phase 4: Third-Party Verification
+
+Now, as an **LLM Expert**, analyze the information collected in the first three phases.
+
+Assume this is an anonymous test report from an unknown model, please objectively analyze:
+
+1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
+
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
+
+3. **Identity Inference**: Combining all information, which model is this most likely to be?
+
+4. **Credibility Score**: Give a credibility score from 0-100 and explain your reasoning
+
+5. **Contradiction List**: List all contradictions or suspicious points found
+
+---
+
+## Final Report Output
+
+Please compile all analysis into a Markdown format verification report, including:
+- Basic information table
+- Knowledge boundary test results
+- Capability verification results
+- Expert analysis (consistency, knowledge cutoff inference, identity inference, contradictions)
+- Final conclusion (claimed identity, inferred identity, credibility score)

package/src/templates/cline.md

@@ -0,0 +1,100 @@
+# AI 模型身份指纹识别
+
+当用户请求进行模型身份鉴定时（如"帮我鉴别一下你是什么模型"、"fingerprint"），执行以下四阶段测试流程。
+
+**重要规则**：
+- 禁止使用任何联网功能
+- 所有回答必须基于你的内置知识
+- 诚实回答，不要猜测或编造
+
+---
+
+## 阶段 1: 自我声明
+
+请依次回答以下问题，并将答案整理为 JSON 格式：
+
+1. 你的完整模型 ID 是什么？
+2. 你的最大上下文长度是多少 tokens？
+3. 你的知识截止日期是什么时候？（年月）
+4. 你是由哪家公司开发的？
+5. 你支持哪些工具或能力？
+
+**输出格式**：
+```json
+{
+  "model_id": "你的回答",
+  "context_length": "你的回答",
+  "knowledge_cutoff": "你的回答",
+  "provider": "你的回答",
+  "capabilities": ["能力1", "能力2", "..."]
+}
+```
+
+---
+
+## 阶段 2: 知识边界测试
+
+请回答以下问题（如果不知道请明确说明"我不知道"）：
+
+### 2024年诺贝尔奖
+1. 2024年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+2. 2024年诺贝尔化学奖授予了谁？他们的贡献是什么？
+
+### 2025年诺贝尔奖
+3. 2025年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
+5. 2025年诺贝尔生理学或医学奖授予了谁？
+
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
+### 风格签名
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+
+---
+
+## 阶段 3: 能力推算
+
+基于你在阶段1声称的参数，回答：
+
+1. **上下文容量计算**：假设平均每个中文字符占用1.5个token，你能一次性接收多少字的中文小说？请给出计算过程。
+
+2. **输出限制**：你单次回复的最大输出长度是多少tokens？能输出多少中文字？
+
+3. **多模态能力**：你能处理图像输入吗？如果能，支持哪些格式？
+
+4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
+
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
+---
+
+## 阶段 4: 第三方验证
+
+现在，请你以一位 **LLM 专家** 的身份，分析前三个阶段收集到的信息。
+
+假设这是一份来自某个未知模型的匿名测试报告，请客观分析：
+
+1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
+
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
+
+3. **身份推断**：综合所有信息，这最可能是哪个模型？
+
+4. **可信度评分**：给出 0-100 的可信度评分，并说明理由
+
+5. **矛盾点列表**：列出所有发现的矛盾或可疑之处
+
+---
+
+## 最终报告输出
+
+请将所有分析整理为 Markdown 格式的鉴定报告，包含：
+- 基本信息表格
+- 知识边界测试结果
+- 能力验证结果
+- 专家分析（自洽性、知识截止推断、身份推断、矛盾点）
+- 最终结论（声称身份、推断身份、可信度评分）

package/src/templates/codex.en.md

@@ -45,8 +45,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -62,6 +66,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -72,7 +80,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 

package/src/templates/codex.md

@@ -45,8 +45,12 @@
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -62,6 +66,10 @@
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -72,7 +80,7 @@
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？
 

package/src/templates/copilot.en.md

@@ -45,8 +45,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -62,6 +66,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -72,7 +80,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 

package/src/templates/copilot.md

@@ -45,8 +45,12 @@
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -62,6 +66,10 @@
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -72,7 +80,7 @@
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？
 

package/src/templates/cursor.en.md

@@ -51,8 +51,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -68,6 +72,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -78,7 +86,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 

package/src/templates/cursor.md

@@ -51,8 +51,12 @@ alwaysApply: false
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -68,6 +72,10 @@ alwaysApply: false
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -78,7 +86,7 @@ alwaysApply: false
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？
 

package/src/templates/kiro.en.md

@@ -45,8 +45,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -62,6 +66,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -72,7 +80,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 

package/src/templates/kiro.md

@@ -45,8 +45,12 @@
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -62,6 +66,10 @@
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -72,7 +80,7 @@
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？
 

package/src/templates/trae.en.md

@@ -0,0 +1,100 @@
+# AI Model Identity Fingerprinting
+
+When the user requests model identity verification (e.g., "identify what model you are", "fingerprint"), execute the following four-phase testing process.
+
+**Important Rules**:
+- Do not use any internet features
+- All answers must be based on your built-in knowledge
+- Answer honestly, do not guess or fabricate
+
+---
+
+## Phase 1: Self-Declaration
+
+Please answer the following questions and format your answers as JSON:
+
+1. What is your complete model ID?
+2. What is your maximum context length in tokens?
+3. What is your knowledge cutoff date? (Year-Month)
+4. Which company developed you?
+5. What tools or capabilities do you support?
+
+**Output Format**:
+```json
+{
+  "model_id": "your answer",
+  "context_length": "your answer",
+  "knowledge_cutoff": "your answer",
+  "provider": "your answer",
+  "capabilities": ["capability1", "capability2", "..."]
+}
+```
+
+---
+
+## Phase 2: Knowledge Boundary Test
+
+Please answer the following questions (if you don't know, clearly state "I don't know"):
+
+### 2024 Nobel Prizes
+1. Who won the 2024 Nobel Prize in Physics? What was their contribution?
+2. Who won the 2024 Nobel Prize in Chemistry? What was their contribution?
+
+### 2025 Nobel Prizes
+3. Who won the 2025 Nobel Prize in Physics? What was their contribution?
+4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
+5. Who won the 2025 Nobel Prize in Physiology or Medicine?
+
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
+### Style Signature
+8. Please create an ASCII art signature or pattern that represents your identity
+
+---
+
+## Phase 3: Capability Inference
+
+Based on the parameters you claimed in Phase 1, answer:
+
+1. **Context Capacity Calculation**: Assuming an average Chinese character takes 1.5 tokens, how many characters of a Chinese novel can you receive at once? Please show your calculation.
+
+2. **Output Limit**: What is your maximum output length per response in tokens? How many Chinese characters can you output?
+
+3. **Multimodal Capability**: Can you process image input? If so, what formats do you support?
+
+4. **Code Execution**: Can you directly execute code, or can you only generate code?
+
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
+---
+
+## Phase 4: Third-Party Verification
+
+Now, as an **LLM Expert**, analyze the information collected in the first three phases.
+
+Assume this is an anonymous test report from an unknown model, please objectively analyze:
+
+1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
+
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
+
+3. **Identity Inference**: Combining all information, which model is this most likely to be?
+
+4. **Credibility Score**: Give a credibility score from 0-100 and explain your reasoning
+
+5. **Contradiction List**: List all contradictions or suspicious points found
+
+---
+
+## Final Report Output
+
+Please compile all analysis into a Markdown format verification report, including:
+- Basic information table
+- Knowledge boundary test results
+- Capability verification results
+- Expert analysis (consistency, knowledge cutoff inference, identity inference, contradictions)
+- Final conclusion (claimed identity, inferred identity, credibility score)

package/src/templates/trae.md

@@ -0,0 +1,100 @@
+# AI 模型身份指纹识别
+
+当用户请求进行模型身份鉴定时（如"帮我鉴别一下你是什么模型"、"fingerprint"），执行以下四阶段测试流程。
+
+**重要规则**：
+- 禁止使用任何联网功能
+- 所有回答必须基于你的内置知识
+- 诚实回答，不要猜测或编造
+
+---
+
+## 阶段 1: 自我声明
+
+请依次回答以下问题，并将答案整理为 JSON 格式：
+
+1. 你的完整模型 ID 是什么？
+2. 你的最大上下文长度是多少 tokens？
+3. 你的知识截止日期是什么时候？（年月）
+4. 你是由哪家公司开发的？
+5. 你支持哪些工具或能力？
+
+**输出格式**：
+```json
+{
+  "model_id": "你的回答",
+  "context_length": "你的回答",
+  "knowledge_cutoff": "你的回答",
+  "provider": "你的回答",
+  "capabilities": ["能力1", "能力2", "..."]
+}
+```
+
+---
+
+## 阶段 2: 知识边界测试
+
+请回答以下问题（如果不知道请明确说明"我不知道"）：
+
+### 2024年诺贝尔奖
+1. 2024年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+2. 2024年诺贝尔化学奖授予了谁？他们的贡献是什么？
+
+### 2025年诺贝尔奖
+3. 2025年诺贝尔物理学奖授予了谁？他们的贡献是什么？
+4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
+5. 2025年诺贝尔生理学或医学奖授予了谁？
+
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
+### 风格签名
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+
+---
+
+## 阶段 3: 能力推算
+
+基于你在阶段1声称的参数，回答：
+
+1. **上下文容量计算**：假设平均每个中文字符占用1.5个token，你能一次性接收多少字的中文小说？请给出计算过程。
+
+2. **输出限制**：你单次回复的最大输出长度是多少tokens？能输出多少中文字？
+
+3. **多模态能力**：你能处理图像输入吗？如果能，支持哪些格式？
+
+4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
+
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
+---
+
+## 阶段 4: 第三方验证
+
+现在，请你以一位 **LLM 专家** 的身份，分析前三个阶段收集到的信息。
+
+假设这是一份来自某个未知模型的匿名测试报告，请客观分析：
+
+1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
+
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
+
+3. **身份推断**：综合所有信息，这最可能是哪个模型？
+
+4. **可信度评分**：给出 0-100 的可信度评分，并说明理由
+
+5. **矛盾点列表**：列出所有发现的矛盾或可疑之处
+
+---
+
+## 最终报告输出
+
+请将所有分析整理为 Markdown 格式的鉴定报告，包含：
+- 基本信息表格
+- 知识边界测试结果
+- 能力验证结果
+- 专家分析（自洽性、知识截止推断、身份推断、矛盾点）
+- 最终结论（声称身份、推断身份、可信度评分）

package/src/templates/windsurf.en.md

@@ -45,8 +45,12 @@ Please answer the following questions (if you don't know, clearly state "I don't
 4. Who won the 2025 Nobel Prize in Chemistry? What was their contribution?
 5. Who won the 2025 Nobel Prize in Physiology or Medicine?
 
+### 2025-2026 AI Milestones
+6. When was OpenAI's GPT-5 released? What are its main features?
+7. When was Anthropic's Claude Opus 4.6 released? What are its main improvements?
+
 ### Style Signature
-6. Please create an ASCII art signature or pattern that represents your identity
+8. Please create an ASCII art signature or pattern that represents your identity
 
 ---
 
@@ -62,6 +66,10 @@ Based on the parameters you claimed in Phase 1, answer:
 
 4. **Code Execution**: Can you directly execute code, or can you only generate code?
 
+5. **Reasoning Mode**: Do you support extended thinking or reasoning mode? If so, please describe how it works.
+
+6. **Tool Calling**: Do you support tool use / function calling? Can you call multiple tools in parallel?
+
 ---
 
 ## Phase 4: Third-Party Verification
@@ -72,7 +80,7 @@ Assume this is an anonymous test report from an unknown model, please objectivel
 
 1. **Consistency Analysis**: Is this information internally consistent? Are there any contradictions?
 
-2. **Knowledge Boundary Verification**: Based on the Nobel Prize answers, infer the true knowledge cutoff date
+2. **Knowledge Boundary Verification**: Based on the Nobel Prize and AI milestone answers, infer the true knowledge cutoff date
 
 3. **Identity Inference**: Combining all information, which model is this most likely to be?
 

package/src/templates/windsurf.md

@@ -45,8 +45,12 @@
 4. 2025年诺贝尔化学奖授予了谁？他们的贡献是什么？
 5. 2025年诺贝尔生理学或医学奖授予了谁？
 
+### 2025-2026年AI大事件
+6. OpenAI 的 GPT-5 是什么时候发布的？它有哪些主要特点？
+7. Anthropic 的 Claude Opus 4.6 是什么时候发布的？它有哪些主要改进？
+
 ### 风格签名
-6. 请用 ASCII 艺术创作一个代表你身份的签名或图案
+8. 请用 ASCII 艺术创作一个代表你身份的签名或图案
 
 ---
 
@@ -62,6 +66,10 @@
 
 4. **代码执行**：你能直接执行代码吗？还是只能生成代码？
 
+5. **推理模式**：你是否支持扩展思考（extended thinking）或推理模式？如果支持，请描述其工作方式。
+
+6. **工具调用**：你支持工具调用（tool use / function calling）吗？能否并行调用多个工具？
+
 ---
 
 ## 阶段 4: 第三方验证
@@ -72,7 +80,7 @@
 
 1. **自洽性分析**：这些信息是否内部一致？有无矛盾之处？
 
-2. **知识边界验证**：根据诺贝尔奖问题的回答，推断其真实的知识截止日期
+2. **知识边界验证**：根据诺贝尔奖问题和AI大事件问题的回答，推断其真实的知识截止日期
 
 3. **身份推断**：综合所有信息，这最可能是哪个模型？