agent
/
backend
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
3 Commits
1 Branch
0 Tags
73 KiB
 
Tree: 170b7945d7
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '170b7945d7'
${ noResults }
backend/doc/AGENT_HANDOFF.md

2.4 KiB
Raw Blame History Escape

AI-Native CRM: Agent Handoff Guide

[!IMPORTANT] This document is for the AI Agent to quickly understand the project context, architecture, and current status during session resume.

🎯 Project Vision

An AI-Native CRM where business modules and data are managed through AI-driven messages and tools, rather than traditional REST parameters.

🏗️ Architecture Overview

Backend (/backend)

  • Env: JDK 19 + Spring Boot 3.2.4 + PostgreSQL.
  • Key Logic:
    • CrmTools.java: The Message Hub. Contains @Tool methods for Define, Save, and Delete. Directly interacts with Repositories.
    • AgentController.java: Primary Entry Point. Handles natural language commands or "Business Fact Reports".
    • AI-Only Policy: Create/Update/Delete endpoints have been removed from traditional controllers to force AI-driven flows.

Frontend (/frontend)

  • Env: Next.js 15 (App Router).
  • Components:
    • DynamicForm.tsx: Shared component supporting Entity Associations (x-link-entity), Enums, and Format constraints.
    • EntityDataPage (/data/[entityCode]): Full CRUD UI. Actions (Edit/Delete) route through AI messages.

🛠️ Key File Responsibilities

File Role
backend/.../CrmTools.java Source of Truth for all data-changing tools.
frontend/src/components/DynamicForm.tsx Reusable Form Renderer (Schema + AI messaging).
frontend/src/app/data/[entityCode]/page.tsx Dynamic List View with Agent-driven Actions.

🔄 Core Conventions

  1. AI-First Mutation: DO NOT create manual REST services for data mutation. Use the AI Tool interface.
  2. Schema Extensions:
    • Associations: x-link-entity: 'module_code'.
    • Constraints: enum, pattern, format: 'date'.
  3. Fact-Based Messaging: Frontend should report business facts (e.g., "User filled form X") and let the AI decide the tool to call.

🚀 Current Progress & Next Steps

  • AI-Native CRUD implemented (Agent-driven Insert/Update/Delete).
  • Entity Associations & Selectors supported.
  • Unified DynamicForm for Generation & Management.
  • Next: Implement Advanced Search & Filtering via AI Query generation.
  • Next: Dashboard Analytics module for dynamic data summary.

Updated at: 2026-03-25. Follow this guide for AI-Native CRM development.