📚 Prompting Basics

Effective prompting is the key to getting the best results from AI models. Here are the fundamental principles:

Core Principles

Be Specific: Clear, detailed requests yield better results
Provide Context: Give the AI background information
Set Expectations: Define the format and style you want
Iterate: Refine your prompts based on responses

Prompt Structure


[Context] + [Task] + [Format] + [Constraints]

🎯 Advanced Techniques

Chain of Thought Prompting

Ask the AI to think step-by-step for complex problems:

Example: "Solve this step by step, showing your reasoning at each stage..."

Role-Based Prompting

Assign a specific role or persona to the AI:

Example: "Act as a senior software engineer and review this code..."

Few-Shot Learning

Provide examples of the desired input-output format:

Example: "Here are 2 examples of good summaries: [examples]. Now summarize this text in the same style..."

🦙 Ollama-Specific Tips

Model Selection

Llama 3.2: Great for general conversation and reasoning
Code Llama: Optimized for programming tasks
Mistral: Excellent for multilingual tasks
Phi-3: Fast responses, good for quick tasks

Context Management

Ollama models have context limits. For long conversations:

Summarize previous context when needed
Use ReactorAI's context selection feature
Start new sessions for different topics

💎 Gemini-Specific Tips

Gemini Model Variants

Gemini Pro: Best for complex reasoning and analysis
Gemini Flash: Faster responses for simpler tasks

Leveraging Gemini's Strengths

Multimodal: Combine text with images for richer context
Large Context: Can handle very long documents
Code Understanding: Excellent at code analysis and generation

⚙️ Session Prompts (System Messages)

Use ReactorAI's session prompt feature to set consistent behavior:

General Assistant


You are a helpful, accurate, and concise assistant. Always provide clear explanations and ask for clarification when needed.

Code Review Assistant


You are an expert code reviewer. Focus on code quality, security, performance, and best practices. Provide specific, actionable feedback.

Creative Writing Partner


You are a creative writing assistant. Help brainstorm ideas, improve prose, and provide constructive feedback on writing style and structure.

💡 Example Prompts

Code Generation

Prompt:

Create a Python function that validates email addresses. Include error handling, type hints, and docstrings. Test it with both valid and invalid examples.

Data Analysis

Prompt:

Analyze this sales data and identify the top 3 trends. Present your findings in a clear format with actionable recommendations: [paste data]

Learning Assistant

Prompt:

Explain quantum computing to me like I'm a computer science student. Use analogies, provide examples, and suggest resources for deeper learning.

🔧 Troubleshooting Common Issues

Getting Inconsistent Results

Make your prompts more specific
Use session prompts for consistent behavior
Provide more context and examples

Responses Too Long/Short

Specify desired length explicitly
Use phrases like "briefly explain" or "provide a detailed analysis"
Give examples of the desired length

AI Not Following Instructions

Break complex requests into smaller steps
Use numbered lists for multi-part instructions
Emphasize critical requirements