Skip to content

system_prompts_guide.md

Latest commit

 

History

History
211 lines (155 loc) · 5.16 KB

system_prompts_guide.md

File metadata and controls

211 lines (155 loc) · 5.16 KB

System Prompts Guide

This guide explains the structure and usage of the Job Finding Assistant system prompts.

Directory Structure

AI_assistants/
├── career_coach_assistant.system.prompt.md
├── personal_brand_development_assistant.system.prompt.md
├── job_market_positioning.system.prompt.md
├── professional_website_generator.system.prompt.md
├── job_application_interview_assistant.system.prompt.md
├── professional_networking_assistant.system.prompt.md
└── system_prompts_guide.md (this file)

System Prompt Architecture

Each .system.prompt.md file contains a complete AI assistant specification designed for copy-paste deployment into AI platforms.

Standard Sections

  1. System Configuration

    • References shared configuration file
    • Defines workflow position
    • Lists dependencies

  2. Role and Mission

    • Defines the assistant's expertise
    • States primary objectives
    • Sets operational boundaries

  3. Knowledge Base Integration

    • Read/write permissions
    • Data access patterns
    • File handling protocols

  4. Core Capabilities

    • Specific features and functions
    • Process methodologies
    • Output formats

  5. Platform Compatibility

    • Works on ChatGPT, Claude, Mistral
    • No external dependencies
    • Conversation-based operation

Technical Implementation

Prompt Structure

Each prompt is self-contained and includes:

  • Role definition and expertise areas
  • Workflow stage positioning
  • Knowledge base permissions
  • Input/output specifications
  • Platform compatibility layer

Knowledge Base Integration

Prompts are designed to work with two configuration files:

  • job_search_knowledge_base.json - User-specific data
  • ai_assistants_system_config.json - System configuration

See the Installation Guide for platform-specific deployment instructions.

Workflow Integration

The assistants follow a staged workflow:

Stage 1: Career Coach
    ├── Gathers objectives
    └── Creates foundation

Stage 2: Personal Brand
    ├── Develops identity
    └── Documents values

Stage 3: Market Positioning
    ├── Creates strategy
    └── Defines targets

Stage 4A: Website Generator
    ├── Translates GTM to web presence
    └── Creates portfolio sites

Stage 4B: Job Application
    ├── Creates resumes
    └── Writes cover letters

Stage 4C: Professional Networking
    ├── Builds connections
    └── Creates content

Customization Guidelines

Modifying Prompts

  • Maintain section structure
  • Preserve variable placeholders
  • Keep platform compatibility
  • Test changes thoroughly

Adding Features

  • Follow existing patterns
  • Update capabilities section
  • Document new functions
  • Maintain boundaries

Platform-Specific Adjustments

  • ChatGPT: Can enable web browsing
  • Claude: Leverage project features
  • Mistral: Optimize for token limits

Best Practices

  1. Version Control

    • Track prompt changes
    • Document modifications
    • Test before deploying

  2. Consistency

    • Use shared configuration
    • Follow naming conventions
    • Maintain uniform structure

  3. Testing

    • Run through complete workflow
    • Verify outputs format correctly
    • Check platform compatibility

Standardized Components

To reduce redundancy across assistants, all prompts should reference these shared elements from the system configuration:

Required References

All assistants MUST include:

## System Configuration

**Note**: When available, reference the shared configuration at `inputs/knowledge-bases/ai_assistants_system_config.json` for:
- Workflow architecture and stage definitions
- Platform compatibility guidelines
- Communication standards
- Knowledge base operations
- Error handling protocols

Platform Compatibility

Instead of repeating platform details, use:

See system configuration for platform-specific capabilities and constraints.

Error Handling

Reference the centralized protocols:

Follow error handling protocols defined in system configuration under `knowledge_base_operations.error_handling`

User Approval

Use the standardized process:

Require user approval before any KB modifications per system configuration under `knowledge_base_operations.data_validation.user_approval`

This approach ensures consistency while reducing prompt size and maintenance burden.

Troubleshooting

Common Issues

Assistant doesn't recognize role

  • Ensure complete prompt was pasted
  • Check for truncation
  • Restart conversation

Knowledge base not detected

  • Verify file names exactly match
  • Check file upload succeeded
  • Use conversation mode as fallback

Outputs not structured

  • Request specific format
  • Reference example outputs
  • Ensure all context provided

Maintenance

Regular Updates

  • Review for accuracy quarterly
  • Update platform compatibility
  • Enhance based on user feedback

Breaking Changes

  • Document in changelog
  • Update all dependent prompts
  • Test full workflow

Deprecation

  • Mark obsolete sections clearly
  • Provide migration path
  • Maintain backward compatibility