///▙▖▙▖▞▞▙▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂
▛//▞▞ ⟦⎊⟧ :: DYNAMIC.PROMPT :: v1.0.0 ▞▞
     Redis-Backed AI Prompt Management
     〔 Prompts in Memory, Not History 〕
///▙▖▙▖▞▞▙▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂

Dynamic Prompt

Store AI prompts in Redis, not conversation history. Modify personality on-the-fly. Save 95% of your tokens.

Traditional AI systems waste thousands of tokens repeating the same system prompt every turn. Dynamic Prompt stores it once in Redis and lets you modify it in real-time with natural language commands.

dp = DynamicPrompt.new
dp.load('assistant_prompt.md')

# Use in any AI call
system_prompt = dp.get_active  # ← From Redis, not repeated

# User says "be more sarcastic" mid-conversation
dp.modify('change tone to sarcastic')  # ← Takes effect immediately

---

Quick Start

gem install dynamic_prompt

require 'dynamic_prompt'

# Initialize
dp = DynamicPrompt.new(redis_url: 'redis://localhost:6379/0')

# Load prompt
dp.load('my_prompt.md')

# Get active prompt (use as system message)
system_prompt = dp.get_active

# Modify on-the-fly
dp.modify('be more concise', user_id: 'user123')

# See changes
puts dp.diff
puts dp.history

---

Features

• 🔄 Load from anywhere - Files, URLs, or direct strings
• ✏️ Natural language mods - "change tone to X", "add rule: Y", "be more Z"
• 📊 Full audit trail - Track who changed what and when
• 🔙 Revert anytime - Restore to original with one command
• 🚀 Multiple formats - Ruby gem, MCP server, Docker container
• 🎯 Provider agnostic - Works with OpenAI, Claude, any LLM

---

The Problem

Traditional approach:
Turn 1:  [1000 token prompt] + user + AI response
Turn 2:  [1000 token prompt] + user + AI response
Turn 20: [1000 token prompt] + user + AI response

Result: 20,000 tokens wasted on repeated prompts

The Solution

Dynamic Prompt approach:
Redis: [prompt:active] ← stored once

Turn 1-20: reference prompt + user + AI response

Result: ~0 tokens (prompt in Redis, not conversation)
Savings: 20,000 tokens per 20-turn conversation

---

Integration Examples

OpenAI

require 'dynamic_prompt'
require 'openai'

dp = DynamicPrompt.new
dp.load('assistant_prompt.md')

client = OpenAI::Client.new(access_token: ENV['OPENAI_API_KEY'])

response = client.chat(
  parameters: {
    model: 'gpt-4',
    messages: [
      { role: 'system', content: dp.get_active },  # ← From Redis
      { role: 'user', content: 'Hello!' }
    ]
  }
)

# Modify mid-conversation
dp.modify('be more concise')
# Next call uses updated prompt automatically

Anthropic Claude

require 'dynamic_prompt'
require 'anthropic'

dp = DynamicPrompt.new
dp.load('assistant_prompt.md')

client = Anthropic::Client.new(api_key: ENV['ANTHROPIC_API_KEY'])

response = client.messages.create(
  model: 'claude-3-opus-20240229',
  system: dp.get_active,  # ← From Redis
  messages: [{ role: 'user', content: 'Hello!' }]
)

---

MCP Server (Cursor Integration)

# Start MCP server
ruby mcp/prompt_server.rb

# Or via Docker
docker-compose up -d

Send JSON commands via stdio:

{"method": "load_prompt", "params": {"source": "/path/to/prompt.md"}}
{"method": "modify_prompt", "params": {"instruction": "be more sarcastic"}}
{"method": "get_active_prompt"}

Available methods: load_prompt, get_active_prompt, modify_prompt, revert_prompt, show_diff, get_history, health_check

---

Docker Deployment

# Quick start
docker-compose up -d

# Or pull from Docker Hub
docker pull yourusername/dynamic-prompt:latest
docker run -e REDIS_URL=redis://your-redis:6379/0 yourusername/dynamic-prompt

Place prompts in ./prompts/ - they mount to /app/prompts/ in container.

---

Natural Language Modifications

Command	Effect
"change tone to sarcastic"	Updates tone
"add rule: never use emojis"	Adds new rule
"remove rule about emojis"	Deletes rule
"default 2 sentences"	Sets response length
"make more verbose"	Increases detail
"be more concise"	Reduces verbosity

---

Architecture

Your AI App (OpenAI/Claude)
         ↓ dp.get_active()
   DynamicPrompt Gem ↔ Redis
         ↑           (prompt:active)
    User commands    (prompt:backup)
    "be more X"      (changelog)

Benefits:

• 95%+ token savings - Prompt stored once, not repeated
• Real-time changes - Modify without restarting
• Full history - Audit trail of all modifications
• Portable - Works with any LLM provider

---

Documentation

• Full Documentation - Complete API reference and examples
• Deployment Guide - Production deployment instructions
• Examples - Working code samples
• Docker Guide - Container deployment

---

Requirements

• Ruby >= 3.0.0
• Redis >= 5.0

---

License

MIT License - See LICENSE.txt

---

Contributing

1. Fork the repo
2. Create feature branch (git checkout -b feature/amazing)
3. Commit changes (git commit -m 'Add feature')
4. Push branch (git push origin feature/amazing)
5. Open Pull Request

---

///▙▖▙▖▞▞▙▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂
  
  SEAL :: DYNAMIC.PROMPT :: v1.0.0
  ⧗ Prompts in Memory, Not History
  ρ{Load} φ{Modify} ν{Evolve}
  
  Ruby Gem • MCP Server • Docker
  95%+ Token Savings
  Credit: LLMDesign 2025
///▙▖▙▖▞▞▙▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂