RubyLLM::Template

Organize prompts into easy-to-use templates for RubyLLM.

# prompts/
#   extract_metadata/
#     ├── system.txt.erb    # System message
#     ├── user.txt.erb      # User prompt  
#     ├── assistant.txt.erb # Assistant message (optional)
#     └── schema.rb         # RubyLLM::Schema definition (optional)

chat = RubyLLM.chat
chat.with_template(:extract_metadata, document: @document).complete

Features

• 🎯 Organized Templates: Structure your prompts in folders with separate files for system, user, assistant, and schema messages
• 🔄 ERB Templating: Use full ERB power with context variables and Ruby logic
• ⚙️ Configurable: Set custom template directories per environment
• 🚀 Rails Integration: Seamless Rails integration with generators and automatic configuration
• 🧪 Well Tested: Comprehensive test suite ensuring reliability
• 📦 Minimal Dependencies: Only depends on RubyLLM and standard Ruby libraries

Installation

Add this line to your application's Gemfile:

gem 'ruby_llm'
gem 'ruby_llm-template'
gem 'ruby_llm-schema'  # Optional for schema.rb support

And then execute:

bundle install

Rails Setup

If you're using Rails, run the generator to set up the template system:

rails generate ruby_llm_template:install

This will:

• Create config/initializers/ruby_llm_template.rb
• Create app/prompts/ directory
• Generate an example template at app/prompts/extract_metadata/

Quick Start

1. Create a Template

Create a directory structure like this:

prompts/
  extract_metadata/
    ├── system.txt.erb    # System message
    ├── user.txt.erb      # User prompt  
    ├── assistant.txt.erb # Assistant message (optional)
    └── schema.rb         # RubyLLM::Schema definition (optional)

2. Write Your Templates

prompts/extract_metadata/system.txt.erb:

You are an expert document analyzer. Extract metadata from documents in a structured format.

prompts/extract_metadata/user.txt.erb:

Please analyze this document: <%= document %>

<% if additional_context %>
Additional context: <%= additional_context %>
<% end %>

prompts/extract_metadata/schema.rb:

# Using RubyLLM::Schema DSL for clean, type-safe schemas
RubyLLM::Schema.create do
  string :title, description: "Document title"
  array :topics, description: "Main topics" do
    string
  end
  string :summary, description: "Brief summary"
  
  # Optional fields with validation
  number :confidence, required: false, minimum: 0, maximum: 1
  
  # Nested objects
  object :metadata, required: false do
    string :author
    string :date, format: "date"
  end
end

3. Use the Template

chat = RubyLLM.chat
chat.with_template(:extract_metadata, document: @document, additional_context: "Focus on technical details").complete

# Under the hood, RubyLLM::Template renders the templates with the context variables
# and applies them to the chat instance using native RubyLLM methods:

# chat.add_message(:system, rendered_system_message)
# chat.add_message(:user, rendered_user_message)
# chat.add_schema(instantiated_schema)

Configuration

Non-Rails Applications

RubyLLM::Template.configure do |config|
  config.template_directory = "/path/to/your/prompts"
end

Rails Applications

The gem automatically configures itself to use Rails.root.join("app", "prompts"), but you can override this in config/initializers/ruby_llm_template.rb:

RubyLLM::Template.configure do |config|
  config.template_directory = Rails.root.join("app", "ai_prompts")
end

Template Structure

Each template is a directory containing ERB files for different message roles:

• system.txt.erb - System message that sets the AI's behavior
• user.txt.erb - User message/prompt
• assistant.txt.erb - Pre-filled assistant message (optional)
• schema.rb - RubyLLM::Schema definition for structured output (optional)

Templates are processed in order: system → user → assistant → schema

ERB Context

All context variables passed to with_template are available in your ERB templates:

chat = RubyLLM.chat
chat.with_template(:message, name: "Alice", urgent: true, documents: @documents)

<!-- /prompts/message/user.txt.erb -->
Hello <%= name %>!

<% if urgent %>
🚨 URGENT: <%= message %>
<% else %>
📋 Regular: <%= message %>
<% end %>

Processing <%= documents.length %> documents:
<% documents.each_with_index do |doc, i| %>
  <%= i + 1 %>. <%= doc.title %>
<% end %>

Advanced Usage

Complex Templates

# Template with conditional logic and loops
RubyLLM.chat.with_template(:analyze_reports,
  reports: @reports,
  priority: "high",
  include_charts: true,
  deadline: 1.week.from_now
).complete

Multiple Template Calls

chat = RubyLLM.chat
  .with_template(:initialize_session, user: current_user)
  .with_template(:load_context, project: @project)
  
# Add more messages dynamically
chat.ask("What should we focus on first?")

Schema Definition with RubyLLM::Schema

The gem integrates with RubyLLM::Schema to provide a clean Ruby DSL for defining JSON schemas. Use schema.rb files instead of JSON:

# prompts/analyze_results/schema.rb
RubyLLM::Schema.create do
  number :confidence, minimum: 0, maximum: 1, description: "Analysis confidence"
  
  array :results, description: "Analysis results" do
    object do
      string :item, description: "Result item"
      number :score, minimum: 0, maximum: 100, description: "Item score"
      
      # Context variables are available
      string :category, enum: categories if defined?(categories)
    end
  end
  
  # Optional nested structures
  object :metadata, required: false do
    string :model_version
    string :timestamp, format: "date-time"
  end
end

Benefits of schema.rb:

• 🎯 Type-safe: Ruby DSL with built-in validation
• 🔄 Dynamic: Access template context variables for conditional schemas
• 📝 Readable: Clean, self-documenting Ruby syntax vs JSON
• 🔧 Flexible: Generate schemas based on runtime conditions
• 🚀 No JSON: Eliminate error-prone JSON string manipulation

Schema-Only Approach: The gem exclusively supports schema.rb files with RubyLLM::Schema. If you have a schema.rb file but the gem isn't installed, you'll get a clear error message.

Error Handling

The gem provides clear error messages for common issues:

begin
  RubyLLM.chat.with_template(:extract_metadata).complete
rescue RubyLLM::Template::Error => e
  puts e.message 
  # "Template 'extract_metadata' not found in /path/to/prompts"
  # "Schema file 'extract_metadata/schema.rb' found but RubyLLM::Schema gem is not installed"
end

Development

After checking out the repo, run bin/setup to install dependencies. You can also run bin/console for an interactive prompt that will allow you to experiment.

To run the test suite:

bundle exec rspec

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/danielfriis/ruby_llm-template.

License

The gem is available as open source under the terms of the MIT License.