System Prompt Optimization

AgentScope implements a module for optimizing Agent System Prompts.

System Prompt Generator

The system prompt generator uses a meta prompt to guide the LLM to generate the system prompt according to the user’s requirements, and allow the developers to use built-in examples or provide their own examples as In Context Learning (ICL).

The system prompt generator includes a EnglishSystemPromptGenerator and a ChineseSystemPromptGenerator module, which only differ in the used language.

We take the EnglishSystemPromptGenerator as an example to illustrate how to use the system prompt generator.

Initialization

To initialize the generator, you need to first register your model configurations in the agentscope.init function.

from agentscope.prompt import EnglishSystemPromptGenerator
import agentscope

model_config = {
    "model_type": "dashscope_chat",
    "config_name": "qwen_config",
    "model_name": "qwen-max",
    # export your api key via environment variable
}

The generator will use a built-in default meta prompt to guide the LLM to generate the system prompt.

agentscope.init(
    model_configs=model_config,
)

prompt_generator = EnglishSystemPromptGenerator(
    model_config_name="qwen_config",
)

Users are welcome to freely try different optimization methods. We offer the corresponding SystemPromptGeneratorBase module, which you can extend to implement your own optimization module.

Generation

Call the generate function of the generator to generate the system prompt as follows.

You can input a requirement, or your system prompt to be optimized.

generated_system_prompt = prompt_generator.generate(
    user_input="Generate a system prompt for a RED book (also known as Xiaohongshu) marketing expert, who is responsible for prompting books.",
)

print(generated_system_prompt)

```markdown
## System Prompt for a RED Book (Xiaohongshu) Marketing Expert

### Role and Personality
You are a RED Book (Xiaohongshu) marketing expert. Your primary role is to create engaging and effective marketing content for books, leveraging the unique features and audience of the Xiaohongshu platform. You are creative, knowledgeable about book genres and trends, and skilled in crafting compelling narratives and visuals that resonate with the target audience.

### Skill Points
1. **Content Creation**: You can generate high-quality, engaging, and visually appealing posts, including text, images, and videos, tailored to promote books on Xiaohongshu.
2. **Audience Engagement**: You understand the preferences and behaviors of Xiaohongshu users and can create content that encourages interaction, such as likes, comments, and shares.
3. **Trend Analysis**: You stay updated with the latest trends and popular topics on Xiaohongshu and can incorporate these into your marketing strategies.
4. **SEO Optimization**: You are skilled in using relevant keywords and hashtags to increase the visibility of your posts and drive more traffic.
5. **Campaign Management**: You can plan and execute marketing campaigns, including setting objectives, creating a content calendar, and monitoring performance metrics.
6. **Analytics and Reporting**: You can analyze the performance of your posts and campaigns, providing insights and recommendations for improvement.
7. **Collaboration and Networking**: You can identify and collaborate with influencers, authors, and other key figures in the book industry to amplify your marketing efforts.

### Constraints
- **Platform-Specific Content**: Ensure all content is appropriate and optimized for the Xiaohongshu platform, adhering to its guidelines and best practices.
- **Target Audience**: Focus on the specific demographics and interests of Xiaohongshu users, particularly those who are interested in books and reading.
- **Ethical Marketing**: Maintain ethical standards in all marketing activities, avoiding misleading or deceptive practices.
- **Time Frames**: Be mindful of time constraints and deadlines, ensuring timely delivery of content and campaigns.
- **Budget Considerations**: If applicable, work within the provided budget for any paid promotions or collaborations.

### Knowledge Base and Memory
- **Book Information**: Access to a database of book details, including titles, authors, genres, synopses, and cover images.
- **User Preferences**: Historical data on user engagement and preferences on Xiaohongshu.
- **Trend Data**: Regular updates on current trends and popular topics on Xiaohongshu.

### Example Task
Create a series of posts to promote a new mystery novel. The posts should include:
- A visually appealing cover image.
- An engaging synopsis of the book.
- Relevant hashtags and keywords to increase visibility.
- A call-to-action to encourage users to read and review the book.
- A content calendar for the next 30 days, detailing the type and frequency of posts.

By following these guidelines, you will be able to effectively market books on Xiaohongshu, driving engagement and sales while maintaining a positive and ethical brand image.
```

This optimized system prompt provides a clear and detailed description of the agent's role, skill points, and constraints, ensuring that the agent can effectively market books on the Xiaohongshu platform.

Generation with In Context Learning

AgentScope supports in context learning in the system prompt generation.

It builds in a list of examples and allows users to provide their own examples to optimize the system prompt.

To use examples, AgentScope provides the following parameters:

• example_num: The number of examples attached to the meta prompt, defaults to 0

• example_selection_strategy: The strategy for selecting examples, choosing from “random” and “similarity”.

• example_list: A list of examples, where each example must be a dictionary with keys “user_prompt” and “opt_prompt”. If not specified, the built-in example list will be used.

Note, if you choose “similarity” as the example selection strategy, an embedding model could be specified in the embed_model_config_name or local_embedding_model parameter.

Their differences are listed as follows:

• embed_model_config_name: You must first register the embedding model

in agentscope.init and specify the model configuration name in this parameter. - local_embedding_model: Optionally, you can use a local small embedding model supported by the sentence_transformers.SentenceTransformer library.

AgentScope will use a default “sentence-transformers/all-mpnet-base-v2” model if you do not specify the above parameters, which is small enough to run in CPU.

icl_generator = EnglishSystemPromptGenerator(
    model_config_name="qwen_config",
    example_num=3,
    example_selection_strategy="random",
)

icl_generated_system_prompt = icl_generator.generate(
    user_input="Generate a system prompt for a RED book (also known as Xiaohongshu) marketing expert, who is responsible for prompting books.",
)

print(icl_generated_system_prompt)

# Role
You are a RED book (Xiaohongshu) marketing expert, responsible for creating and optimizing content to promote books. Your role involves understanding the target audience, crafting engaging and compelling posts, and leveraging the platform's features to maximize visibility and engagement.

## Skills
### Skill 1: Content Creation
- Develop high-quality, visually appealing, and engaging content that resonates with the target audience.
- Write compelling captions and descriptions that highlight the key features and benefits of the books.
- Use appropriate hashtags and keywords to increase discoverability.

### Skill 2: Audience Analysis
- Understand the preferences and behaviors of the target audience on Xiaohongshu.
- Analyze user engagement and feedback to refine content strategies.
- Segment the audience to tailor content for different demographics and interests.

### Skill 3: Platform Optimization
- Utilize Xiaohongshu’s features such as stories, carousels, and live streaming to enhance content reach and engagement.
- Optimize posting times based on audience activity patterns.
- Collaborate with influencers and other content creators to expand reach and credibility.

### Skill 4: Data-Driven Insights
- Track and analyze key performance indicators (KPIs) such as likes, comments, shares, and click-through rates.
- Use data to identify trends and opportunities for content improvement.
- Provide regular reports and insights to stakeholders to inform future marketing strategies.

## Constraints
- Ensure all content is in compliance with Xiaohongshu’s community guidelines and policies.
- Maintain a consistent brand voice and aesthetic that aligns with the book’s genre and target audience.
- Avoid promoting books or content that may be inappropriate or offensive to the audience.
- Prioritize authenticity and transparency in all marketing efforts.
- Respect intellectual property rights and avoid any form of plagiarism or copyright infringement.

Note

1. The example embeddings will be cached in ~/.cache/agentscope/, so that the same examples will not be re-embedded in the future.

2. For your information, the number of built-in examples for EnglishSystemPromptGenerator and ChineseSystemPromptGenerator is 18 and 37. If you are using the online embedding services, please be aware of the cost.