apidash/doc/proposals/2025/fosshack/mavericks_chatbot.md

Add ChatBot for API Assistance

Issue - https://github.com/foss42/apidash/issues/605
PR - https://github.com/foss42/apidash/pull/608
Link - https://fossunited.org/hack/fosshack25/p/7e6upj6f19

What we planned to do:

We aim to implement a ChatBot feature in API Dash with the following capabilities:

1. Explain API:
   • Description: Analyze API responses and provide explanations in natural language.
   • Features: Generate clear descriptions of the API, including:
     • Purpose of the API.
     • Key parameters and their roles.
     • Response structure and fields.
   • Useful for documentation and understanding API behavior.
2. Debug Requests Based on Status Codes & Error Messages:
   • Description: Provide structured debugging suggestions for failed API requests.
   • Features:
     • Analyze status codes (e.g., 4xx, 5xx) and error messages.
     • Offer step-by-step guidance to resolve issues.
3. Generate Test Cases:
   • Description: Automatically generate test cases for API endpoints.
   • Features: Includes
     • Valid input scenarios.
     • Edge cases (e.g., invalid inputs, boundary values).
     • Expected responses and status codes.
   • Ensure test cases are ready to use in testing frameworks
4. Generate Sample Codes:
   • Description: Generate ready-to-run code snippets for various programming languages (e.g., Dart, Python, JavaScript, React, Flutter).
   • Features: Includes
     • API integration code.
     • Error handling.
     • UI components for frontend frameworks (e.g: React).
   • Ensure the code is directly usable with minimal modifications.

What we built

1. Ollama Model Integration

• Dependency: We integrated the AI chatbot with API Dash using the ollama_dart: ^0.2.2 package.
• System Requirements: To use the Ollama model, the system must have the Ollama application installed locally with a compatible model.
• Implementation:
  • The interaction with the Ollama model is handled in lib/services/ollama_services.dart.
  • We tested multiple models, including deepseek:r1:1.5b, ollama3.2:3b, and llama3.2:1b.
  • Among these, ollama3.2:3b delivered the most accurate responses.
  • Recommendation: Use ollama3.2:3b or higher models for better accuracy.
  • Customization: Developers can switch models by modifying the “model:''” parameter in /apidash/lib/services/ollama_service.dart.

2. Dynamic UI: ChatBot Widget

• Design:
  • A mini widget (?) is placed at the bottom right of the response panel.
  • When clicked, it expands into an AI chatbot interface.
• Features:
  • Explain API: Provides explanations for API functionalities.
  • Debug API: Assists in debugging API-related issues.
  • Test Cases: Generates test cases for APIs.
  • Generate Code: Helps in generating code snippets for API integration.
  • General Questions: Users can ask general questions about ApiDash.
• Markdown Support:
  • We used the “package:flutter_markdown/flutter_markdown.dart”package to format and display responses in a clean, readable manner.

Setup Guide

• Install the updated dependencyollama_dart: ^0.2.2
• To use the Ollama model, the system must have the Ollama application installed locally with a compatible model.
• The interaction with the Ollama model is handled in lib/services/ollama_services.dart.
• Recommendation: Use ollama3.2:3b or higher models for better accuracy.
• Developers need to switch models by modifying the “model:''” parameter in /apidash/lib/services/ollama_service.dart to their local Ollama model to interact.
• There you go now run flutter run to use AI Chatbot in ApiDash.

Issues

1. Enhance Response Formatting: Improve the AI to ensure consistent markdown formatting in responses, eliminating plain text outputs.
2. Optimize Structured Outputs: Fine-tune the AI to generate more organized and structured responses for better usability and clarity.

Future Enhancements:

1. Enable users to integrate their local Ollama models seamlessly through the ChatBot interface for enhanced customization.
2. Enable one-click downloading and copying of test codes, ensuring a seamless workflow for developers.
3. Implementing a feature to generate test codes dynamically based on API endpoints and user inputs.
4. A feature to Integrate with complete system architecture
5. Suggesting the best practices from selected type of requests.