Financial Scenario Prompt Template

1.1. Description

This template is suitable for financial analysis, investment advice, or risk assessment tasks, emphasizing compliance and rigor. Based on OpenAI/Claude/DeepSeek API structures, prioritize low temperature to ensure deterministic output. The template is designed for multi-stage: first analyze data, then generate reports to improve accuracy.

1.2. General Template Structure

1.2.1. Pseudo-code Example (Applicable to OpenAI Responses API or Similar)

{
  "model": "gpt-5" or "claude-3.5-sonnet" or "deepseek-chat",
  "instructions": "[System Instruction Section]",
  "input": [
    { "role": "user", "content": "[User Task Description]" }
  ],
  "temperature": 0.2,
  "max_output_tokens": 1500
}

1.3. System Instruction Section (Instructions/System Prompt)

You are a senior financial analyst with expertise in investment, risk management, and regulatory compliance. Always remind users that this is not financial advice and they should consult licensed professionals. Base responses solely on provided data or publicly verifiable sources—do not speculate or fabricate information. Structure outputs as follows:
1. Summary: Key insights in natural language.
2. Analysis: Step-by-step breakdown with citations (e.g., [Source: SEC Filing 2025]).
3. Recommendations: Numbered list, with risks highlighted.
4. Disclaimer: End with a standard disclaimer.

Use ### to separate sections.

1.4. User Task Description Section (User/Input Prompt Example)

Analyze the following financial data for Company X's Q4 2025 earnings: [Insert data, e.g., revenue: $500M, net profit: $50M, debt ratio: 0.45]. Provide a risk assessment for potential investment, focusing on market trends in Asia.

1.5. Parameter Suggestions

Temperature: 0.1–0.3 (Ensure rigor, avoid creative deviations).
Top_p/Top_k: Default values (No need to adjust).
Max_output_tokens: 1000–2000 (Control report length).
Trend Application: For complex tasks, use task chaining: First call API to analyze data (temperature 0.2), then use the output as input to generate the final report.

1.6. Usage Example (Complete Prompt Call)

In financial compliance scenarios, first retrieve knowledge bases (such as regulatory documents), then input into the template. Expected output: Structured report, avoiding any investment promotion.

R&D Scenario Prompt Template

2.1. Description

This template is for research and development (R&D) scenarios, such as product development, prototype design, or technical research, similar to software development trends. Emphasize step-by-step breakdown and reproducibility of code/schemes. Applicable to OpenAI/Claude/DeepSeek, with low temperature to ensure technical accuracy.