LangChain4j & Model Context Protocol (MCP) Integration

SyncBoot integrates LangChain4j as its enterprise AI framework and exposes Model Context Protocol (MCP) interfaces to communicate with external orchestrators and AI clients.

---

MCP Integration Architecture

SyncBoot provides standardized Tools and Resources via HTTP Server-Sent Events (SSE).

---

Default MCP Tool Registry

Tool Name	Description	Parameters	Return Type
syncboot_read_schema	Inspects table structure and column metadata	tableName (String)	JSON (Columns, Types, Nullability, Comments)
syncboot_execute_query	Executes authorized domain CRUD queries	sql (String), params (Map)	Record set (max 100 rows)
syncboot_fetch_server_logs	Extracts recent cluster error logs	lines (Int), level (ERROR)	Stack trace text
syncboot_get_table_list	Retrieves all registered business domain tables	tenantId (String)	Array of table names and descriptions

---

Java Backend Tool Implementation (LangChain4j)

@Component
public class SyncBootMcpTools {

    @Autowired
    private SchemaService schemaService;

    @Autowired
    private LogInspectorService logInspectorService;

    @Tool("Retrieves DDL and metadata for specified table.")
    public TableMetaDTO readSchema(
            @P("Physical table name (e.g. TB_ORDER)") String tableName) {
        return schemaService.getTableMetadata(tableName);
    }

    @Tool("Fetches recent server error logs for analysis.")
    public String fetchServerLogs(
            @P("Number of log lines (default 100)") int lines) {
        return logInspectorService.getRecentErrorLogs(lines);
    }
}