llm_conductor 0.1.1 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c1b006030ac01906369f830f2c11d958f40c3bf3a48404179d78c6b1e75a82e5
-  data.tar.gz: 7ccf09702230675a381f164f4ccb91034aec880c5769cb8887711c6737385c46
+  metadata.gz: c692236c5d26b598a2275a3a6591022312aec4e4a4d4ee57a186abc774bc39ce
+  data.tar.gz: 3c20edc85442b8e5227fb6896d3899f08495fa1296306042c07c8bc62a883941
 SHA512:
-  metadata.gz: 4c3283602c064e5c7e67ac5a1b26353c936b9d5d351ad21d87a89058aa045af1f8f916f350ac07db4a0405a6fe9091b09ecc00b091afd57039d3ae560bd199e8
-  data.tar.gz: 571400d81fa8029ecf9f9d4a53c48ac26392c13de927994c358b0e953d27257b94ff929f1d65f13b64ccddaada8785e33f271b4cf99ebdad74c63abef3dd4a97
+  metadata.gz: 8ed23461382c3a8ee2fddea05fdaef5c72111bac8e4b820e4310a40105a8b0bd78b8f1c36c9176f9ac7ddf8c7b4300e9208b45711a129ab0531dd579f9a52964
+  data.tar.gz: d446efb23252acee7ea453361d402089b87c56450e20ed7e45b1aab2cd0381bff442ed51ea7e2a52a544c39f3117795ad17ac2208dc12514f211d12ba150bb2c

data/.rubocop.yml

@@ -2,8 +2,6 @@
 plugins:
   - rubocop-performance
   - rubocop-rake
-
-require: 
   - rubocop-capybara
   - rubocop-factory_bot
   - rubocop-rspec
@@ -27,21 +25,14 @@ Style/StringLiteralsInInterpolation:
 
 Style/HashSyntax:
   EnforcedShorthandSyntax: always
-# This is not rails application.
-# Rails/Blank:
-#   Enabled: false
-
-# Rails/Present:
-#   Enabled: false
-
-# Rails/TimeZone:
-#   Enabled: false
 
 Lint/ConstantDefinitionInBlock:
   Enabled: false
 
 Metrics/MethodLength:
   Max: 15
+  Exclude:
+    - 'lib/llm_conductor/prompts.rb'
 
 RSpec/ExampleLength:
   Enabled: false
@@ -67,7 +58,7 @@ RSpec/MultipleDescribes:
 RSpec/SpecFilePathFormat:
   Enabled: false
 
-RSpec/FilePath:
+RSpec/SpecFilePathSuffix:
   Enabled: false
 
 RSpec/UnspecifiedException:
@@ -94,6 +85,19 @@ Metrics/BlockLength:
   Exclude:
     - '*.gemspec'
 
+# Prompt template methods naturally have high complexity due to conditional string building
+Metrics/AbcSize:
+  Exclude:
+    - 'lib/llm_conductor/prompts.rb'
+
+Metrics/CyclomaticComplexity:
+  Exclude:
+    - 'lib/llm_conductor/prompts.rb'
+
+Metrics/PerceivedComplexity:
+  Exclude:
+    - 'lib/llm_conductor/prompts.rb'
+
 Layout/LineLength:
   Max: 120
 

data/README.md

@@ -52,14 +52,18 @@ puts response.estimated_cost   # Cost in USD
 ### 2. Template-Based Generation
 
 ```ruby
-# Use built-in templates with structured data
+# Use built-in text summarization template
 response = LlmConductor.generate(
   model: 'gpt-5-mini',
-  type: :summarize_description,
+  type: :summarize_text,
   data: {
-    name: 'Ekohe',
-    domain_name: 'ekohe.com',
-    description: 'An AI company specializing in...'
+    text: 'Ekohe (ee-koh-hee) means "boundless possibility." Our way is to make AI practical, achievable, and most importantly, useful for you — and we prove it every day. With almost 16 years of wins under our belt, a market-leading 24-hr design & development cycle, and 5 offices in the most vibrant cities in the world, we surf the seas of innovation. We create efficient, elegant, and scalable digital products — delivering the right interactive solutions to achieve your audience and business goals. We help you transform. We break new ground across the globe — from AI and ML automation that drives the enterprise, to innovative customer experiences and mobile apps for startups. Our special sauce is the care, curiosity, and dedication we offer to solve for your needs. We focus on your success and deliver the most impactful experiences in the most efficient manner. Our clients tell us we partner with them in a trusted and capable way, driving the right design and technical choices.',
+    max_length: '20 words',
+    style: 'professional and engaging',
+    focus_areas: ['core business', 'expertise', 'target market'],
+    audience: 'potential investors',
+    include_key_points: true,
+    output_format: 'paragraph'
   }
 )
 
@@ -67,7 +71,7 @@ response = LlmConductor.generate(
 if response.success?
   puts "Generated: #{response.output}"
   puts "Tokens: #{response.total_tokens}"
-  puts "Cost: $#{response.estimated_cost}"
+  puts "Cost: $#{response.estimated_cost || 'N/A (free model)'}"
 else
   puts "Error: #{response.metadata[:error]}"
 end
@@ -143,20 +147,6 @@ response = LlmConductor.generate(
 )
 ```
 
-**Supported Claude Models:**
-- `claude-3-5-sonnet-20241022` (Latest Claude 3.5 Sonnet)
-- `claude-3-5-haiku-20241022` (Claude 3.5 Haiku)
-- `claude-3-opus-20240229` (Claude 3 Opus)
-- `claude-3-sonnet-20240229` (Claude 3 Sonnet)
-- `claude-3-haiku-20240307` (Claude 3 Haiku)
-
-**Why Choose Claude?**
-- **Superior Reasoning**: Excellent for complex analysis and problem-solving
-- **Code Generation**: Outstanding performance for programming tasks
-- **Long Context**: Support for large documents and conversations
-- **Safety**: Built with safety and helpfulness in mind
-- **Cost Effective**: Competitive pricing for high-quality outputs
-
 ### Google Gemini (Automatic for Gemini models)
 ```ruby
 response = LlmConductor.generate(
@@ -172,18 +162,6 @@ response = LlmConductor.generate(
 )
 ```
 
-**Supported Gemini Models:**
-- `gemini-2.5-flash` (Latest Gemini 2.5 Flash)
-- `gemini-2.5-flash` (Gemini 2.5 Flash)
-- `gemini-2.0-flash` (Gemini 2.0 Flash)
-
-**Why Choose Gemini?**
-- **Multimodal**: Native support for text, images, and other modalities
-- **Long Context**: Massive context windows for large documents
-- **Fast Performance**: Optimized for speed and efficiency
-- **Google Integration**: Seamless integration with Google services
-- **Competitive Pricing**: Cost-effective for high-volume usage
-
 ### Ollama (Default for non-GPT/Claude/Gemini models)
 ```ruby
 response = LlmConductor.generate(

data/lib/llm_conductor/prompts.rb

@@ -1,124 +1,140 @@
 # frozen_string_literal: true
 
 module LlmConductor
-  # Collection of pre-built prompt templates for common LLM tasks including
-  # content analysis, link extraction, and data summarization.
+  # Collection of general-purpose prompt templates for common LLM tasks
   module Prompts
-    def prompt_featured_links(data)
+    # General prompt for extracting links from HTML content
+    # More flexible and applicable to various use cases
+    def prompt_extract_links(data)
+      criteria = data[:criteria] || 'relevant and useful'
+      max_links = data[:max_links] || 10
+      link_types = data[:link_types] || %w[navigation content footer]
+
       <<~PROMPT
-        You are an AI assistant tasked with analyzing a webpage's HTML content to extract the most valuable links. Your goal is to identify links related to features, products, solutions, pricing, and social media profiles, prioritizing those from the same domain as the current page. Here are your instructions:
+        Analyze the provided HTML content and extract links based on the specified criteria.
+
+        HTML Content:
+        #{data[:html_content] || data[:htmls]}
+
+        Extraction Criteria: #{criteria}
+        Maximum Links: #{max_links}
+        Link Types to Consider: #{link_types.join(', ')}
 
-        - You will be provided with the HTML content of the current page in the following format:
-        <page_html>
-        #{data[:htmls]}
-        </page_html>
+        #{"Domain Filter: Only include links from domain #{data[:domain_filter]}" if data[:domain_filter]}
 
-        - Parse the HTML content and extract all hyperlinks (a href attributes). Pay special attention to links in the navigation menu, footer, and main content areas.
+        Instructions:
+        1. Parse the HTML content and identify all hyperlinks
+        2. Filter links based on the provided criteria
+        3. Prioritize links from specified areas: #{link_types.join(', ')}
+        4. Return up to #{max_links} most relevant links
+        #{if data[:format] == :json
+            '5. Format output as a JSON array of URLs'
+          else
+            '5. Format output as a newline-separated list of URLs'
+          end}
 
-        - Filter and prioritize the extracted links based on the following criteria:
-           a. The link must be from the same domain as the current URL.
-           b. Prioritize links containing keywords such as "features", "products", "solutions", "pricing", "about", "contact", or similar variations.
-           c. Include social media profile links (e.g., LinkedIn, Instagram, Twitter, Facebook) if available.
-           d. Exclude links to login pages, search pages, or other utility pages.
+        Provide only the links without additional commentary.
+      PROMPT
+    end
 
-        - Select the top 3 most valuable links based on the above criteria.
+    # General prompt for content analysis and data extraction
+    # Flexible template for various content analysis tasks
+    def prompt_analyze_content(data)
+      content_type = data[:content_type] || 'webpage content'
+      analysis_fields = data[:fields] || %w[summary key_points entities]
+      output_format = data[:output_format] || 'structured text'
 
-        - Format your output as a JSON array of strings, where each string is a full URL. Use the following format:
-        <output_format>
-        ["https://example.com/about-us", "https://example.com/products", "https://example.com/services"]
-        </output_format>
+      <<~PROMPT
+        Analyze the provided #{content_type} and extract the requested information.
 
-        - The links must be the same domain of following
-        <domain>
-          #{data[:current_url]}
-        </domain>
+        Content:
+        #{data[:content] || data[:htmls] || data[:text]}
 
-        If fewer than 3 relevant links are found, include only the available links in the output array.
+        Analysis Fields:
+        #{analysis_fields.map { |field| "- #{field}" }.join("\n")}
 
-        Remember to use the full URL for each link, including the domain name. If you encounter relative URLs, combine them with the domain from the current URL to create absolute URLs.
+        #{"Additional Instructions:\n#{data[:instructions]}" if data[:instructions]}
 
-        Provide your final output without any additional explanation or commentary.
+        #{if output_format == 'json'
+            json_structure = analysis_fields.map { |field| "  \"#{field}\": \"value or array\"" }.join(",\n")
+            "Output Format: JSON with the following structure:\n{\n#{json_structure}\n}"
+          else
+            "Output Format: #{output_format}"
+          end}
+
+        #{"Constraints:\n#{data[:constraints]}" if data[:constraints]}
+
+        Provide a comprehensive analysis focusing on the requested fields.
       PROMPT
     end
 
-    def prompt_summarize_htmls(data)
+    # General prompt for text summarization
+    # Applicable to various types of text content
+    def prompt_summarize_text(data)
+      max_length = data[:max_length] || '200 words'
+      focus_areas = data[:focus_areas] || []
+      style = data[:style] || 'concise and informative'
+
       <<~PROMPT
-        Extract useful information from the webpage including a domain, detailed description of what the company does, founding year, country, business model, product description and features, customers and partners, development stage, and social media links. output will be json
-
-        You are tasked with extracting useful information about a company from a given webpage content. Your goal is to analyze the content and extract specific details about the company, its products, and its operations.
-
-        You will be provided with raw HTML content in the following format:
-
-        <html_content>
-        #{data[:htmls]}
-        </html_content>
-
-        Carefully read through the webpage content and extract the following information about the company:
-
-        - Name(field name): The company's name
-        - Domain name(field domain_name): The company's domain
-        - Description(field description): A comprehensive explanation of what the company does
-        - Country(field country): The company's country
-        - Region(field region): The company's region
-        - Location(field location): The company's location
-        - Founding on(field founded_on): Which year the company was established
-        - Business model(field business_model): How the company generates revenue
-        - Product description(product_description): A brief overview of the company's main product(s) or service(s)
-        - Product features(product_features): Key features or capabilities of the product(s) or service(s)
-        - Customers and partners(field customers_and_partners): Notable clients or business partners
-        - Development stage(field development_stage): The current phase of the company (e.g., startup, growth, established)
-        - Social media links(field social_media_links): URLs to the company's social media profiles
-          - instagram_url
-          - linkedin_url
-          - twitter_url
-
-        If any of the above information is not available in the webpage content, use "Not available" as the value for that field.
-
-        Present your findings in a JSON format. Here's an example of the expected structure:
-
-        <output_format>
-        {
-          "name": "AI-powered customer service",
-          "domain_name": "example.com",
-          "description": "XYZ Company develops AI chatbots that help businesses automate customer support...",
-          "founding_on": 2018,
-          "country": "United States",
-          "Region": "SA",
-          "Location": "SFO",
-          "business_model": "SaaS subscription",
-          "product_description": "AI-powered chatbot platform for customer service automation",
-          "product_features": ["Natural language processing", "Multi-language support", "Integration with CRM systems"],
-          "customers_and_partners": ["ABC Corp", "123 Industries", "Big Tech Co."],
-          "development_stage": "Growth",
-          "social_media_links": {
-            "linkedin_url": "https://www.linkedin.com/company/xyzcompany",
-            "twitter_url": "https://twitter.com/xyzcompany",
-            "instagram_url": "https://www.instagram.com/xyzcompany"
-          }
-        }
-        </output_format>
-
-        Remember to use only the information provided in the webpage content. Do not include any external information or make assumptions beyond what is explicitly stated or strongly implied in the given content.
-
-        Present your final output in JSON format, enclosed within <json_output> tags.
+        Summarize the following text content.
+
+        Text:
+        #{data[:text] || data[:content] || data[:description]}
+
+        Summary Requirements:
+        - Maximum Length: #{max_length}
+        - Style: #{style}
+        #{"- Focus Areas: #{focus_areas.join(', ')}" if focus_areas.any?}
+        #{"- Target Audience: #{data[:audience]}" if data[:audience]}
+
+        #{'Include key points and main themes.' if data[:include_key_points]}
+
+        #{if data[:output_format] == 'bullet_points'
+            'Format as bullet points.'
+          elsif data[:output_format] == 'paragraph'
+            'Format as a single paragraph.'
+          end}
+
+        Provide a clear and accurate summary.
       PROMPT
     end
 
-    def prompt_summarize_description(data)
+    # General prompt for data classification and categorization
+    # Useful for various classification tasks
+    def prompt_classify_content(data)
+      categories = data[:categories] || []
+      classification_type = data[:classification_type] || 'content'
+      confidence_scores = data[:include_confidence] || false
+
       <<~PROMPT
-        Given the company's name, domain, description, and a list of industry-related keywords,
-        please summarize the company's core business and identify the three most relevant industries.
-        Highlight the company's unique value proposition, its primary market focus,
-        and any distinguishing features that set it apart within the identified industries.
-        Be as objective as possible.
-
-        Name: #{data[:name]}
-        Domain Name: #{data[:domain_name]}
-        Industry: #{data[:industries]}
-        Description: #{data[:description]}
+        Classify the provided #{classification_type} into the most appropriate category.
+
+        Content to Classify:
+        #{data[:content] || data[:text] || data[:description]}
+
+        Available Categories:
+        #{categories.map.with_index(1) { |cat, i| "#{i}. #{cat}" }.join("\n")}
+
+        #{"Classification Criteria:\n#{data[:classification_criteria]}" if data[:classification_criteria]}
+
+        #{if confidence_scores
+            'Output Format: JSON with category and confidence score (0-1)'
+          else
+            'Output Format: Return the most appropriate category name'
+          end}
+
+        #{if data[:multiple_categories]
+            "Note: Multiple categories may apply - select up to #{data[:max_categories] || 3} most relevant."
+          else
+            'Note: Select only the single most appropriate category.'
+          end}
+
+        Provide your classification based on the content analysis.
       PROMPT
     end
 
+    # Flexible custom prompt template
+    # Allows for dynamic prompt creation with variable substitution
     def prompt_custom(data)
       template = data.fetch(:template)
       template % data

data/lib/llm_conductor/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module LlmConductor
-  VERSION = '0.1.1'
+  VERSION = '1.0.0'
 end

data/lib/llm_conductor.rb

@@ -74,9 +74,10 @@ module LlmConductor
 
   # List of supported prompt types
   SUPPORTED_PROMPT_TYPES = %i[
-    featured_links
-    summarize_htmls
-    summarize_description
+    extract_links
+    analyze_content
+    summarize_text
+    classify_content
     custom
   ].freeze
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: llm_conductor
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 1.0.0
 platform: ruby
 authors:
 - Ben Zheng
 bindir: exe
 cert_chain: []
-date: 2025-09-25 00:00:00.000000000 Z
+date: 2025-09-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport