
Who this is for
- Readers who can use chat tools but can't write prompts
- Engineers who need to precipitate a team Prompt specification
What problem does this article solve?
Prompt is written like a temporary chat, and the results are not reusable, reviewable, and maintainable.
Key concepts
- mission instruction
- background context
- constraints
- FEW-SHOT example
- output format
Walkthrough
- Write a bad Prompt sample
- Structured Prompt
- join boundary conditions
- Compare before and after using the same input
Python code example
The companion demo is generated in:
cd examples/llm-guides/005_prompt_instruction_context
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 05 asks what prompt engineering actually tunes. Prompts written as one-off chat messages are hard to reuse, review, or 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, large model, artificial intelligence, AIGC, Prompt
WeChat article summary
Breaking a prompt into its components shows how to turn a one-off question into something a team can reuse.
Alternative title for Zhihu
- Is the Prompt project a metaphysics?
- What parts is a good Prompt made of?
- Why is the same question different from the question?
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.