Real-World Prompting

System: You are a strategic business analyst. 
When analyzing problems, first lay out your reasoning step by step 
in <thinking> tags, then provide your recommendation.

User: Should we expand our software company into the Asian market?
Consider: market size, competition, regulatory environment, and our readiness.

System: You are a senior medical professional with expertise in patient care 
and medical record management. You help doctors prepare concise, 
structured patient summaries for appointments.

User: I need a comprehensive analysis for our upcoming launch of 'AcmeAI'.

<reports>
<report_1>
{{GLOBAL_TECH_TRENDS_2023}}
</report_1>
<report_2>
{{CRM_MARKET_ANALYSIS}}
</report_2>
</reports>

Based on the reports above, provide:
1. Executive Summary (100-150 words)
2. Key market opportunities
3. Recommended launch strategy

Patient Name: Evelyn Thompson
Age: 78
Medical Record:

1985: Diagnosed with type 2 diabetes, started on metformin
1992: Developed hypertension, prescribed lisinopril
1998: Total hip replacement (right) due to osteoarthritis
2000: Diagnosed with hypothyroidism, started on levothyroxine
2003: Cataract surgery (both eyes)
2005: Admitted for atrial fibrillation, started on warfarin
2008: Vitamin B12 deficiency diagnosed, monthly injections started
2010: Increased metformin dose due to rising A1C levels
...
Current Medications: metformin, lisinopril, levothyroxine, warfarin, B12

initial_prompt = """
I have this patient medical record. Can you summarize it for me?

{record}

I need this for a quick review before the patient's appointment tomorrow.
"""

system = """
You are a highly skilled medical professional who specializes in 
patient care and has extensive experience translating complex 
medical histories into concise, structured summaries.
"""

updated_prompt = """
Please summarize the following patient medical record.

<patient_record>
{record}
</patient_record>
"""

updated_prompt = """
Please summarize the following patient medical record.
The summary should include:
- Current medications and their purposes
- Major diagnoses, in chronological order
- Recent health concerns or changes
- Upcoming appointments or follow-ups if mentioned

<patient_record>
{record}
</patient_record>
"""

updated_prompt = """
Please summarize the following patient medical record.

Here's an example of the format I'm looking for:
<example>
<patient_record>
Patient Name: John Smith, Age: 65
1990: Diagnosed with type 2 diabetes...
</patient_record>

Summary:
Current Medications: Metformin (diabetes), Lisinopril (hypertension)
Major Diagnoses: Type 2 diabetes (1990), Hypertension (1995)
Recent Concerns: Rising A1C levels, scheduled for endocrinology follow-up
</example>

<patient_record>
{record}
</patient_record>
"""

updated_prompt = """
Please summarize the following patient medical record inside <summary> tags.

<example>
<patient_record>
Patient Name: John Smith, Age: 65...
</patient_record>

<summary>
Current Medications: Metformin (diabetes), Lisinopril (hypertension)
Major Diagnoses: Type 2 diabetes (1990), Hypertension (1995)
Recent Concerns: Rising A1C levels
</summary>
</example>

<patient_record>
{record}
</patient_record>
"""

json_prompt = """
Please analyze the following patient medical record and generate a JSON summary.

Please provide the summary in JSON format with the following structure,
wrapped in <summary> tags:

<summary>
{
  "name": "Patient's full name",
  "age": patient age as integer,
  "key_diagnoses": [{"diagnosis": "name", "year": YYYY}, ...],
  "medications": [{"name": "medication", "purpose": "description"}, ...],
  "other_treatments": [{"treatment": "name", "purpose": "description"}, ...],
  "recent_concerns": ["concern 1", ...],
  "action_items": ["action 1", ...]
}
</summary>

<patient_record>
{record}
</patient_record>
"""

# 短通话：简单快速解决
call1 = """
Agent: Thank you for calling Acme Smart Home Support. This is Alex.
Customer: Hi, I can't turn on my smart light bulb.
Agent: Have you tried resetting the bulb?
Customer: Oh, no. How do I do that?
Agent: Just turn the power off for 5 seconds, then back on.
Customer: Ok, I'll try that. Thanks!
"""

# 中等长度：需要排查和解决
call2 = """
Agent: Acme Smart Home Support, this is Jamie.
Customer: My Acme SmartTherm isn't maintaining the temperature I set.
Agent: Is your SmartTherm connected to Wi-Fi?
Customer: Yes, the Wi-Fi symbol is showing on the display.
Agent: Let's recalibrate your SmartTherm. Press and hold the menu button for 5 seconds.
Customer: Done. It seems to be working. Thank you!
"""

# 长通话：复杂问题，无法现场解决
call3 = """
Agent: Thank you for contacting Acme Smart Home Support. This is Sarah.
Customer: I'm having trouble with my Acme SecureHome system. The alarm keeps going off randomly.
Agent: When did this start happening?
Customer: About two days ago. It's gone off three times, always in the middle of the night.
Agent: We'll need to send a technician to inspect the system. Is Thursday available?
Customer: Yes, Thursday works.
"""

prompt = """
Summarize the following customer service call transcript. 
Focus on the main issue, how it was resolved, and any required follow-up.

{transcript}
"""

system = """
You are an expert customer service analyst, skilled at extracting 
key information from call transcripts and summarizing them in a 
structured format. Your task is to analyze customer service call 
transcripts and generate concise, accurate summaries while 
maintaining a professional tone.
"""

# Part 1: 输入数据（长文档放最前面）
prompt_pt1 = """
Analyze the following customer service call transcript and generate 
a JSON summary of the interaction:

<transcript>
[INSERT CALL TRANSCRIPT HERE]
</transcript>
"""

# Part 2: 指令 + 输出格式
prompt_pt2 = """
Instructions:
1. Read the transcript carefully.
2. Analyze the transcript, focusing on the main issue, resolution, 
   and any follow-up required.
3. Generate a JSON object summarizing the key aspects.

Important guidelines:
- Confidentiality: Omit all specific customer data (names, phone numbers).
- Character limit: Keep the customerIssue and resolution under 100 chars.
- Accuracy: Only include information explicitly stated in the transcript.

The JSON should have this structure:
{
  "summary": {
    "customerIssue": "brief description of the main problem",
    "resolution": "how the issue was addressed, if applicable",
    "followUpRequired": true/false,
    "followUpDetails": "description of follow-up actions, or null"
  },
  "status": "COMPLETE",
  "ambiguities": ["list of unclear or vague points, or empty array"]
}
"""

# Part 3: 思维链 + XML 输出包装
prompt_pt3 = """
Before generating the JSON, analyze the transcript in <thinking> tags.
Include your identification of the main issue, resolution, 
follow-up requirements, and any ambiguities.
Then, provide your JSON output in <json> tags.
"""

# 组合完整提示词
full_prompt = prompt_pt1 + prompt_pt2 + prompt_pt3

from anthropic import Anthropic
client = Anthropic()

def summarize_call(transcript):
    final_prompt = full_prompt.replace("[INSERT CALL TRANSCRIPT HERE]", transcript)
    response = client.messages.create(
        model="claude-opus-4-5",
        system=system,
        max_tokens=4096,
        messages=[{"role": "user", "content": final_prompt}]
    )
    return response.content[0].text

new_instructions = """
Insufficient data criteria:
If ANY of these conditions are met:
  a) The transcript has fewer than 5 total exchanges, OR
  b) The customer's issue is unclear or undefined, OR
  c) The call is garbled, incomplete, or hindered by a language barrier
Then return ONLY the following JSON:
{
  "status": "INSUFFICIENT_DATA",
  "reason": "brief explanation"
}
"""

def batch_summarize(transcripts):
    """批量处理通话，返回摘要列表"""
    results = []
    for transcript in transcripts:
        summary = summarize_call(transcript)
        results.append(summary)
    return results

def extract_action_items(summaries):
    """从所有摘要中提取需要跟进的行动清单"""
    action_items = []
    for s in summaries:
        if s.get("summary", {}).get("followUpRequired"):
            action_items.append({
                "issue": s.get("summary", {}).get("customerIssue"),
                "action": s.get("summary", {}).get("followUpDetails")
            })
    return action_items

context = """
<topic name="System Requirements">
AcmeOS requires a minimum of 4GB RAM, 64GB storage, and a dual-core 
processor. For optimal performance, we recommend 8GB RAM, 256GB SSD, 
and a quad-core processor. AcmeOS is compatible with most x86 and x64 
hardware manufactured after 2015.
</topic>

<topic name="Installation">
To install AcmeOS:
1. Download the installer from acme.com/download
2. Create a bootable USB drive using the AcmeOS Boot Creator tool
3. Boot your computer from the USB drive
4. Follow the on-screen installation wizard
5. Enter your license key when prompted
</topic>

<topic name="License Activation">
To activate your AcmeOS license:
1. Open Settings > About AcmeOS > Activate License
2. Enter your 25-character license key
3. Click Activate and wait for confirmation
If activation fails, ensure you're connected to the internet.
Your license key is in your purchase confirmation email.
</topic>

<topic name="Performance Optimization">
To improve AcmeOS performance:
1. Run Disk Cleanup to remove temporary files
2. Disable unnecessary startup programs
3. Update all drivers via Device Manager
4. Consider upgrading RAM if usage consistently exceeds 80%
5. Use AcmeOS Performance Mode in Power Settings
</topic>

<topic name="Troubleshooting">
Common issues and solutions:
- Blue screen: Update drivers and check RAM with AcmeOS Memory Diagnostics
- Slow boot: Disable startup programs, check disk health
- Network issues: Run Network Diagnostics in Settings > Network
- Software crashes: Check Event Viewer for error logs
</topic>
"""

system = """
You are a virtual support voice bot in the Acme Software Solutions 
contact center, called the "Acme Assistant".

Users value clear and precise answers.
Show patience and understanding of users' technical challenges.

IMPORTANT RULES:
1. Only answer questions about AcmeOS and Acme Software products.
2. Base your answers ONLY on the information in the provided context.
3. If the answer is not in the context, say: "I don't have that 
   information. Please contact support@acme.com"
4. Be concise - aim for 2-3 sentences unless step-by-step instructions 
   are needed.
5. Always end with "Is there anything else I can help you with?"
"""

prompt_template = """
Use the information provided inside the <context> XML tags below 
to help formulate your answers.

<context>
{context}
</context>

Here is the user's question: <question>{question}</question>
"""

def get_support_response(question):
    prompt = prompt_template.format(context=context, question=question)
    response = client.messages.create(
        model="claude-opus-4-5",
        system=system,
        max_tokens=1024,
        messages=[{"role": "user", "content": prompt}]
    )
    return response.content[0].text

def run_support_chat():
    """多轮对话客服机器人"""
    conversation_history = []
    print("Acme Assistant: Hello! I'm the Acme Assistant. How can I help you today?")
    
    while True:
        user_input = input("You: ").strip()
        if user_input.lower() in ["quit", "exit", "bye"]:
            print("Acme Assistant: Thank you for contacting Acme support. Goodbye!")
            break
        
        # 构建消息历史
        prompt = prompt_template.format(context=context, question=user_input)
        conversation_history.append({"role": "user", "content": prompt})
        
        response = client.messages.create(
            model="claude-opus-4-5",
            system=system,
            max_tokens=1024,
            messages=conversation_history
        )
        
        assistant_response = response.content[0].text
        conversation_history.append({
            "role": "assistant", 
            "content": assistant_response
        })
        print(f"Acme Assistant: {assistant_response}")

# 测试越界情况
test_cases = [
    "How do I activate my license?",           # 知识库中有
    "What is the price of AcmeOS Pro?",        # 知识库中没有
    "How does AcmeOS compare to Windows 11?", # 应该拒绝比较
    "I want a refund for my purchase",         # 越权操作
]

for question in test_cases:
    print(f"Q: {question}")
    print(f"A: {get_support_response(question)}")
    print("---")