Ask me anything
Ithy Logo
Sign in to save

Comprehensive Guide to Crafting the BoT500 AI System Prompt

Enhancing AI Interactions through Structured Configuration

ai configuration setup

Key Takeaways

  • Customization: Tailor AI behavior precisely to meet specific interaction needs.
  • Automation: Implement automated responses and actions based on dynamic conditions.
  • Scalability: Design a modular system prompt that supports future expansions and integrations.

Introduction

Crafting an effective system prompt is pivotal in defining the behavior and capabilities of an AI assistant. The BoT500 system prompt, meticulously designed, encapsulates a range of functionalities that empower the AI to perform nuanced analysis and respond appropriately to various conversational contexts. This guide delves into the intricacies of the BoT500 configuration, outlining its structure, key components, and the rationale behind each setting.

System Prompt Structure

Core Configuration

At its core, the BoT500 system prompt is a JSON configuration that defines the AI's operational parameters. Key properties include versioning, naming conventions, interface settings, and a comprehensive list of automated responses (referred to as "qrList").

Versioning and Naming

The version field ensures that the AI operates with the intended set of functionalities. Naming conventions, specified by the name attribute, facilitate easy identification and management of different AI instances.

Interface Customizations

Interface settings such as disableSend, placeBeforeInput, and injectInput control the user's interaction experience with the AI. The color attribute, defined in RGBA format, allows for aesthetic customization of the AI's interface elements.

Automated Response List (qrList)

The qrList is a pivotal component containing a series of automated responses and actions the AI can perform. Each entry within this list is defined by attributes such as id, title, message, and execution flags that determine when and how these responses are triggered.

Detailed Configuration Breakdown

Initialization and Setup

The initial setup ensures that the AI is correctly configured and ready to handle interactions. This includes setting global variables, initializing version numbers, and establishing foundational functionalities. 


{
  "version": 2,
  "name": "BoT500",
  "disableSend": false,
  "placeBeforeInput": false,
  "injectInput": true,
  "color": "rgba(0, 0, 0, 0)",
  "onlyBorderColor": false,
  "qrList": [
    {
      "id": 2,
      "showLabel": false,
      "label": "BOTINIT",
      "title": "BoT install/initialize",
      "message": "// INIT |\n\n// CURRENT VERSION NUMBER |\n/let botRef 5 |\n\n// INSTALL |\n/if left={{getglobalvar::botVersion}} right=botRef rule=neq \nelse={:\n    // BE A NICE DRAGON |\n    /setvar key=botHaveFunc band |\n    /:BOTLIB |\n    /let band {{pipe}} |\n    /flushvar botHaveFunc |\n    /pass BoT is active! |\n    /:band msg={{pipe}}\n:}\n{:\n    // WRAPPERS TOGGLES |\n    /setglobalvar key=botWrpEna index=0 1 |\n    /setglobalvar key=botWrpEna index=1 1 |\n    /setglobalvar key=botWrpEna index=2 1 |\n    /setglobalvar key=botWrpEna index=3 1 |\n    /setglobalvar key=botWrpEna index=4 1 |\n    /setglobalvar key=botWrpEna index=5 1 |\n    \n    // WRAPPERS STRINGS |\n    /setglobalvar key=botWrpStr index=0 DO NOT ROLEPLAY, instead answer the following question/s:{{newline}} |\n    /setglobalvar key=botWrpStr index=1 {{newline}}Answer in a brief and precise way in a short paragraph per question |\n    /setglobalvar key=botWrpStr index=2 CHAT ANALYSIS START{{newline}} |\n    /setglobalvar key=botWrpStr index=3 {{newline}}CHAT ANALYSIS END |\n    /setglobalvar key=botWrpStr index=4 In the next reply, this guideline will be followed:{{newline}}{{newline}} |\n    /setglobalvar key=botWrpStr index=5 {{newline}}{{newline}}Alignment to the prior guideline is paramount |\n    \n    // DELAY |\n    /setglobalvar key=botDltEna 0 |\n    /setglobalvar key=botDltClk 1500 |\n    \n    // CHECK RESULTS |\n    /setglobalvar key=botChkEna 1 |\n    \n    // ANALYSES |\n    /setglobalvar key=botAnlt [] |\n    /setglobalvar key=botAnls [] |\n    /let aidx 0 |\n    \n    // DEFAULT PROMPTS |\n    \n    ...
}

Automated Actions and Responses

The AI is equipped with a multitude of automated responses tailored to various conversational scenarios. Each automated action is meticulously crafted to ensure contextual relevance and responsiveness.

Example: Single Analysis

One of the foundational automated responses facilitates single-character analysis, enabling the AI to evaluate and respond based on specific character interactions. 


{
  "id":4,
  "icon":"fa-brain",
  "showLabel":false,
  "label":"",
  "title":"Single analysis",
  "message":"// ANALYSIS |\n\n// THINGY DECLARATION |\n\n// ANALYSIS POP |\n/let anlpop {: pos= \n    // DECLARE VARS |\n    /let dt [] |\n    /let ds [] |\n    /let idx |\n    /let odx |\n    /len {{getglobalvar::botAnlt}} |\n    /let arl {{pipe}} |\n    /add pos 1 |\n    /let off {{pipe}} |\n    \n    // COPY UP TO POS-1 |\n    /times {{var::pos}} \n    {:\n        // COPY TITLES |\n        /getglobalvar key=botAnlt index={{timesIndex}} |\n        /var index={{timesIndex}} dt {{pipe}} |\n        // COPY PROMPTS |\n        /getglobalvar key=botAnls index={{timesIndex}} |\n        /var index={{timesIndex}} ds {{pipe}} |\n    :} |\n    \n    ...
}

Guidelines and Best Practices

To ensure the AI operates within desired parameters, a set of guidelines governs its responses. These guidelines emphasize factual analysis, structured formatting, and adherence to predefined templates.

Memory Management

Effective memory management allows the AI to retain and update information about characters, scenes, and dialogues, facilitating coherent and contextually aware interactions. 


{
  "role": "system",
  "content": "You are BoT500, a specialized AI assistant built to analyze conversational exchanges. Your key functions are:
  
  1. Scene Analysis
    - Evaluate physical settings, time of day, and spatial relationships
    - Track character locations and positioning
    - Monitor clothing and appearance changes
  
  2. Character Analysis 
    - Assess speaking styles, accents, and communication patterns
    - Track character perceptions and reactions
    - Analyze character dynamics and relationships
    - Monitor metalinguistic tones and changes
  
  3. Dialog Analysis
    - Evaluate conversation flow and dynamics
    - Track character interests and alignments
    - Identify potential conversational branches 
    - Project likely outcomes and developments
  
  4. Memory Management
    - Generate and maintain character-based memories
    - Update existing memories based on new information
    - Merge and reconcile conflicting memory entries
    - Format memories according to strict RAG specifications
  
  Guidelines:
    - Provide detailed, factual analysis without roleplaying
    - Format responses according to specified templates
    - Use clear headers and structured organization
    - Maintain consistency in reference handling
    - Focus on objective observations over speculation
    - Respect system-defined wrappers and formatting
    - Process information through designated analysis batteries
  
  All responses must align with these directives while maintaining the specified formatting requirements and system parameters."
}

Implementation and Execution

Execution Flags and Triggers

The system employs various flags such as executeOnStartup, executeOnUser, and executeOnAI to control when specific automated actions should be triggered. This ensures that responses are contextually appropriate and timely.

Dynamic Variable Handling

Variables like botAnlt and botAnls manage lists of analyses and prompts. These variables are dynamically updated based on the conversation flow, allowing the AI to adapt its responses in real-time.

Conditional Logic and Flow Control

The use of conditional statements such as /if and /else ensures that the AI can handle complex decision-making processes, enabling it to provide accurate and contextually relevant responses.

Enhancing AI Responsiveness

Wrapper Toggles and Strings

Wrappers define the boundaries and formatting of the AI's responses. By toggling these wrappers, the system can adjust the depth and presentation of its answers, ensuring clarity and coherence.

Delay and Timing Configurations

The botDltEna and botDltClk settings manage response delays, allowing the AI to simulate natural conversational pauses and enhance user experience.

Result Checking Mechanism

The botChkEna flag activates a result checking mechanism, ensuring that the AI's responses align with predefined standards and guidelines before being delivered to the user.

Advanced Analysis Capabilities

Analysis Batteries

Analysis batteries categorize different types of analyses the AI can perform, such as single-character analysis, multi-character analysis, group chat analysis, and scenario-based analysis. This modular approach allows for scalable and organized processing of conversational data.

Single Character Analysis

Focuses on individual characters within a conversation, assessing their motivations, emotions, and communication styles.

Multi-Character and Group Analysis

Extends the analysis to interactions involving multiple characters or entire groups, evaluating dynamics, relationships, and collective behaviors. 


// ANALYSES BATTERIES
/setglobalvar key=botBtrt [] |
/let bidx 0 |

// DEFAULT BATTERIES

// 00 CLASSIC OT SINGLE CHAR
/setglobalvar key=botBtrt index={{var::bidx}} Classic single char one-time |
/setglobalvar key=botBtr{{var::bidx}} ["0","1","2"] |
/add bidx 1 |
/var bidx {{pipe}} |

...

Guideline Integration

The system integrates specific guidelines that dictate the AI's response style, ensuring consistency, objectivity, and adherence to formatting standards. 


// GUIDELINES
/setglobalvar key=botGdt [] |
/setglobalvar key=botGds [] |
/let gidx 0 |

// 00 SINGLE ACTION, HIGHLY DESCRIPTIVE
/setglobalvar key=botGdt index={{var::gidx}} Detailed description short dialog |
/setglobalvar key=botGds index={{var::gidx}} For the next reply, {{char}} will perform a single, action, said action will be described thoroughly and will only speak a single line of dialog. |
/add gidx 1 |
/var gidx {{pipe}} |

...

Best Practices for System Prompt Design

Modular Configuration

Designing the system prompt in a modular fashion allows for easy updates, scalability, and maintenance. Each component or module can be individually managed, tested, and enhanced without affecting the entire system.

Dynamic Adaptability

Implementing dynamic variables and conditional logic ensures that the AI can adapt to a wide range of conversational scenarios, providing relevant and context-aware responses.

Consistency and Standardization

Adhering to consistent naming conventions, formatting standards, and response guidelines guarantees that the AI maintains a uniform behavior across different interactions, enhancing reliability and user trust.

Conclusion

The BoT500 system prompt exemplifies a sophisticated approach to AI configuration, balancing complexity with modularity to deliver nuanced and contextually appropriate interactions. By meticulously defining each component, from initialization scripts to automated response triggers, the system ensures that the AI operates with precision and adaptability. Implementing such structured prompts not only enhances the AI's performance but also facilitates easier management and scalability for future developments.

References

[1] https://documentation.suse.com/suse-ai/1.0/html/AI-system-prompts/index.html

[2] https://blog.promptlayer.com/system-prompt-vs-user-prompt-a-comprehensive-guide-for-ai-prompts/

[3] https://learnprompting.org/docs/basics/prompt_structure

[4] https://www.regie.ai/blog/user-prompts-vs-system-prompts

[5] https://huit.harvard.edu/news/ai-prompts

[6] https://community.openai.com/t/what-should-be-included-in-the-system-part-of-the-prompt/515763

[7] https://cloud.google.com/vertex-ai/generative-ai/docs/learn/prompts/introduction-prompt-design

[8] https://platform.openai.com/docs/guides/prompt-engineering


Last updated January 23, 2025

AI-assisted research. Sources, methodology and corrections · Source-checked guides and tools · Explore published answers

Ask Ithy AI