super-ai-agent

# Introduction A showcase project that demonstrates the true nature of LLM (Large Language Model) development through the implementation of Manus agents, MCP server, MCP Client, offline RAG (Retrieval-Augmented Generation), Function Call, and more. # Environment - Spring AI - SpringBoot 3 - JDK21 # Agent Implementation Effects ### The agent can autonomously call tools from the locally implemented MCP service, think independently, and ultimately provide results.  ### During the execution process, the agent can independently decide which tool to call and determine the next strategy based on the results returned from the tool calls:  ### Generated PDF results (in markdown format, image loading issues):  ### The image link can be opened:  # Completed Features - Custom Interceptor (Advisor) - Prohibited Words Verification Advisor - Dedicated Template Class for loading templates from files - Custom ChatMemory - Structured Output Converter - Conversation Memory Persistence - Local RAG - Online RAG - Vector Database - Document Retriever - Context Query Enhancer - Function Call Development - MCP Development (sse and stdio) - Custom Manus Agent # TODO - Permission Verification - Custom Conversation Memory, persisting conversations to MySQL and Redis - Custom DocumentReader for reading GitHub repository information - Custom QueryTransformer for query transformation, using third-party translation APIs instead of large model translation - Implement hybrid retrieval based on vector databases and other databases (MySQL, Redis, ES) - Automatic email sending tool - Optimize PDF generation, local storage of images - Manual control of ToolCallingManager execution flow - Implement env environment variable parameter passing for multi-agent workflows