

Who this is for
- chatbot developer
- AI application backend engineer
- Corporate knowledge base maintainer
What problem does this article solve?
All rules are tucked into user Prompt, resulting in overpowering, conflict and difficult maintenance.
Key concepts
- system instructions
- developer constraints
- user input
- Priority
- Prompt injection
Walkthrough
- Responsibilities for splitting roles
- Put fixed rules into high priority messages
- Isolate user input separately
- Add conflict detection tips
Python code example
The companion demo is generated in:
cd examples/llm-guides/007_prompt_role_policy
python demo.py
The example follows two principles:
- From
OPENAI_API_KEY, load the key at runtime to avoid hard-coding it. - Without an API key, the example runs in mock mode so you can follow the workflow.
The essential call looks like this:
import os
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
print("未配置 OPENAI_API_KEY,进入 mock 模式。")
else:
# 真实项目中在这里初始化 SDK 客户端并发起请求。
print("已读取 OPENAI_API_KEY,可以调用模型。")
Common pitfalls
- Publishing after a single successful demo run leaves out error handling, logging and human review.
- Including API keys, real file paths, account details or business data in published articles or code.
- Readers can struggle to reproduce an example if it is unclear which parts are mocked and which run in production.
- Citing an API or framework feature without checking its documentation can mislead readers when versions change.
Summary
- LLM Guide 07 explains how to divide responsibilities among system, developer, and user messages. Putting every rule in the user prompt can let instructions exceed their authority, create conflicts, and make rules hard to maintain.
- Make the workflow reproducible, reversible and easy to review. One good generation tells you little about how it will hold up in use.
- Before publishing, check current documentation for model capabilities, API parameters and third-party frameworks.
CSDN tags
Prompt, ChatGPT, large model application, security, software architecture
WeChat article summary
The guide explains how an application can assign instructions to the appropriate message roles to reduce unauthorized overrides and conflicting rules.
Alternative title for Zhihu
- What is the difference between System Prompt and User Prompt?
- Where should chatbot rules be placed?
- Why not write all constraints into user input?
References and further reading
- Check the OpenAI documentation for your model, SDK and account permissions.
- Python documentation on environment variables, virtual environments and exception handling.
- Consult the documentation for the tools used in the example, such as LangChain, LlamaIndex, Haystack or the vector database.