CustomAgent - extends API for code RAG

[phodal]

Generate by ChatGPT

GitHub Issue Roadmap for RebuildWise

Milestone 1: Project Initiation and System Understanding

2. Issue 2: System Analysis with LLM

   • Description: Use LLM to analyze the existing legacy system's source code, data structures, and documentation. Document key findings.
   • Labels: system-analysis, documentation

3. Issue 3: Requirements Gathering with RAG

   • Description: Utilize RAG to retrieve and compile domain-specific knowledge for gathering and documenting user requirements.
   • Labels: requirements, research

Milestone 2: Refactoring Planning

4. Issue 4: Initial Refactoring Plan with LLM

   • Description: Generate an initial refactoring plan using LLM, including high-level architecture design and technology stack selection.
   • Labels: refactoring-plan, architecture

5. Issue 5: Best Practices Research with RAG

   • Description: Use RAG to find information on best practices and case studies related to legacy system refactoring. Integrate these insights into the refactoring plan.
   • Labels: best-practices, research

Milestone 3: Implementation and Testing

6. Issue 6: Code Generation with LLM

   • Description: Employ LLM to generate code for standard tasks and modules, streamlining the development process.
   • Labels: code-generation, development

7. Issue 7: Test Case Generation with LLM

   • Description: Utilize LLM to automatically create test cases and test data, ensuring thorough testing of the refactored system.
   • Labels: testing, quality-assurance

Milestone 4: Documentation and Knowledge Sharing

8. Issue 8: Documentation Enhancement with LLM

   • Description: Use LLM to improve system documentation, making it more comprehensive and user-friendly.
   • Labels: documentation, knowledge-sharing

9. Issue 9: Technical Support with LLM

   • Description: Implement LLM-powered technical support to assist developers in troubleshooting, performance optimization, and security enhancements.
   • Labels: technical-support, development

This roadmap provides a structured plan for the RebuildWise project, with a focus on leveraging LLM and RAG throughout the different stages of system refactoring. You can create these GitHub issues in your repository and assign them to relevant team members, setting priorities and due dates as needed. Remember to adapt the roadmap based on your project's specific requirements and goals.

[phodal]

1. support for Embedding URL like CoUnit?
2. Local embedding model in Code

[phodal]

Scene 1: DirectResponse

First version APIs:

[
  {
    "name": "API  集成",
    "url": "https://rag.unitmesh.cc/api/market",
    "responseAction": "Direct",
    "interactive": "ChatPanel"
  }
]

Default request format in JSON like custom fields? with query?=xxx

Default Response format in JSON Path ?

Scene 2: TextChunk

First version APIs:

[
  {
    "name": "API  集成",
    "url": "https://rag.unitmesh.cc/api/market",
    "responseAction": "TextChunk",
    "interactive": "AppendCursor"
  }
]