# Project export: Jade.AI This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small. ## Project metadata - Hackathon: Cal Hacks 12.0 - Tagline: Cursor for Data Analysis - Devpost: https://devpost.com/software/jadeai - GitHub: https://github.com/erik-ksth/jade-ai.git - Video: https://www.youtube.com/embed/NeEhUii2XYk?enablejsapi=1&hl=en_US&rel=0&start=&version=3&wmode=transparent - Result: winner (Groq: Best Use of Groq) - Team: 2 GitHub contributor(s) — aungbbo (3 commits), erik-ksth (3 commits) ## Devpost submission (written by the team) ### Overview Tired of spending 70% of your time cleaning data? Jade.AI is an intelligent, all-in-one platform that lets you clean, analyze, and visualize your data using simple English commands. No code, no app-switching -- just results. ### Inspiration As a college student passionate about data analysis, I’ve faced a frustrating reality: most of my project time isn't spent on analysis. It's spent on grunt work. I often wasted over 70% of my time just cleaning messy datasets. I was constantly switching between Excel for quick fixes, Pandas for heavy-duty cleaning, SQL for data pulling, and another tool like Tableau for visualization. The process was slow, exhausting, and fragmented. I wanted a single, intelligent, all-in-one solution that could handle all these tasks with just a few natural language commands. I envisioned a "Cursor for data analysis", an AI-first tool that would let me talk to my data. That's why we built Jade.AI. ### What it does Jade.AI empowers anyone, regardless of their technical background, to work with data effortlessly. It’s as simple as 1-2-3: Upload your data (CSV, Excel, etc.). Type what you want in plain English. Get immediate results. You can clean, manipulate, visualize, and summarize, all in one seamless interface: “Clean the entire dataset” → Jade.AI intelligently handles missing values, removes duplicates, fixes inconsistent formatting, and standardizes data types. “Show a pie chart of the gender distribution” → Instantly generates a beautiful, interactive chart with auto-labeled insights. “What’s the average salary by department?” → Returns a clean summary table or a bar chart, your choice. “Summarize the key findings from this data” → Creates a concise, human-readable report. ### How we built it We built Jade.AI to be fast, intelligent, and scalable. Frontend: We built a clean and responsive UI using Next.JS to create a seamless, single-page application experience. Backend: We used a Python (FastAPI) backend to create REST APIs that handles all user requests. LLM: We integrated the Groq API as the main LLM to power our natural language-to-action engine. Its incredible speed makes the conversation feel instantaneous. This model interprets user commands and translates them into executable operations. Data Manipulation: The legendary Pandas library is our workhorse. All data cleaning, transformation, and analysis commands are piped through a secure Pandas runtime. Data Visualization: We used Chart.js to generate rich, interactive, and beautiful visualizations that can be embedded directly in the chat interface. ### Challenges we ran into Prompt Engineering is Deceptive: Getting the LLM to consistently and safely translate a vague command like "fix the messy columns" into the correct sequence of Pandas functions was our biggest hurdle. It required lots of iterative prompt design and building a robust validation layer. Maintaining Data State: Managing the state of a user's DataFrame across multiple, independent API calls was a huge challenge. Our initial implementation was slow and buggy before we used Groq. The Scope Creep: We had so many ideas! We wanted to add SQL database connections, advanced statistical modeling, and more. Focusing on our core loop (Upload → Clean → Analyze → Viz) within the time limit was a real test of discipline. ### Accomplishments we're proud of It Actually Works! We have a fully functional, end-to-end demo. You can upload a genuinely messy dataset and walk away with clean data and insightful charts in under two minutes. The "Wow" Moment: The biggest win was testing it on a friend with zero coding experience. They were able to perform a complete data analysis task that would have taken them hours, and they did it with a smile. The "Cursor for Data" Feel: We successfully captured that "magic" feeling. It feels less like a tool and more like an intelligent collaborator, which was our core vision. ### What we learned LLMs are Interpreters, Not Magicians: The real magic isn't just the LLM; it's the pipeline. The most critical work is building the robust "plumbing" that translates the LLM's intent into safe, executable code. UX is Everything: For a tool aimed at non-technical users, simplicity is the ultimate feature. A single, clear chat box proved to be infinitely more powerful than a complex dashboard. Speed is a Feature: Using an incredibly fast LLM (Groq) was a game-changer. When the analysis feels instant, it encourages creativity and exploration, which is the entire point of data analysis. ### What's next We're just getting started. Our vision is to make data analysis as easy as having a conversation. More Data Connectors: We plan to add support for connecting directly to SQL databases, Google Sheets, and other live APIs. Advanced Analytics: We're working on moving beyond descriptive statistics to include predictive modeling. (e.g., "Forecast next quarter's sales based on this data"). "Analysis History": We want to create a feature that lets users track, revert, and branch their analysis steps, almost like Git for data analysis. ## README (from the GitHub repository) No README available. ## Detected evidence (automated analysis) Indexed codebase: 56 recognized source files, 253 KB. - CSS (language) — detected in the code - FastAPI (technology) — detected in the code - Next.js (technology) — detected in the code - Python (language) — detected in the code - React (technology) — detected in the code - Tailwind CSS (technology) — detected in the code - TypeScript (language) — detected in the code ## Codebase structure (from repository index) ### Files (71 of 71) ``` .DS_Store backend/.gitignore backend/aiAgent.py backend/api/__init__.py backend/api/models.py backend/api/routes/__init__.py backend/api/routes/chat_routes.py backend/api/routes/data_routes.py backend/api/routes/upload_routes.py backend/api/utils.py backend/core/__init__.py backend/core/config.py backend/core/state.py backend/main.py backend/requirements.txt backend/services/__init__.py backend/services/chart_service.py backend/services/dataframe_service.py backend/services/execution_service.py backend/services/file_service.py backend/services/quality_service.py backend/tools/__init__.py backend/tools/code_executor.py backend/tools/dataframe_utils.py backend/tools/snapshot_manager.py backend/vercel.json backend/workflows/__init__.py backend/workflows/cleaning_orchestrator.py backend/workflows/nodes/__init__.py backend/workflows/nodes/code_executor.py backend/workflows/nodes/code_generator.py backend/workflows/nodes/intent_classifier.py backend/workflows/nodes/quality_assessor.py backend/workflows/nodes/response_generator.py backend/workflows/orchestrator.py backend/workflows/README.md backend/workflows/state.py dirtyData/awkward_input.csv dirtyData/Cafe Sales Dirty Data.csv dirtyData/COLA Data ETL.csv dirtyData/Women Concert Tour.csv docs/architecture-discussion.md frontend/.expo/README.md frontend/.expo/settings.json frontend/.gitignore frontend/components.json frontend/eslint.config.mjs frontend/next.config.ts frontend/package.json frontend/postcss.config.mjs frontend/README.md frontend/src/app/globals.css frontend/src/app/layout.tsx frontend/src/app/page.tsx frontend/src/components/ChatAgent.tsx frontend/src/components/Dashboard.tsx frontend/src/components/DataTable.tsx frontend/src/components/FileExplorer.tsx frontend/src/components/SimpleTabs.tsx frontend/src/components/theme-provider.tsx frontend/src/components/ThemeToggle.tsx frontend/src/components/ui/alert-dialog.tsx frontend/src/components/ui/button.tsx frontend/src/components/ui/card.tsx frontend/src/components/ui/dialog.tsx frontend/src/components/ui/dropdown-menu.tsx frontend/src/components/ui/input.tsx frontend/src/components/ui/resizable.tsx frontend/src/lib/utils.ts frontend/tsconfig.json shared/types.ts ``` ### Dependencies - backend/requirements.txt: annotated-types@==0.7.0, anyio@==4.11.0, cachetools@==6.2.0, certifi@==2025.10.5, charset-normalizer@==3.4.3, click@==8.3.0, et_xmlfile@==2.0.0, fastapi@==0.118.0, groq@==0.33.0, h11@==0.14.0, httpcore@==1.0.7, httpx@==0.27.2, idna@==3.10, langchain-core@==0.3.21, langchain-groq@==0.2.1, langgraph@==0.2.45, numpy@==2.3.3, openpyxl@==3.1.5, pandas@==2.3.3, pyasn1@==0.6.1, pyasn1_modules@==0.4.2, pydantic@==2.11.9, pydantic_core@==2.33.2, python-dateutil@==2.9.0.post0, python-dotenv@==1.0.0, python-multipart@==0.0.20, pytz@==2025.2, requests@==2.32.5, rsa@==4.9.1, six@==1.17.0, sniffio@==1.3.1, starlette@==0.48.0, tenacity@==9.1.2, typing_extensions@==4.15.0, typing-inspection@==0.4.2, tzdata@==2025.2, uagents@==0.22.10, urllib3@==2.5.0, uvicorn@==0.37.0, websockets@==15.0.1 - frontend/package.json: @eslint/eslintrc@^3, @radix-ui/react-alert-dialog@^1.1.15, @radix-ui/react-dialog@^1.1.15, @radix-ui/react-dropdown-menu@^2.1.16, @radix-ui/react-slot@^1.2.3, @tailwindcss/postcss@^4, @types/node@^20, @types/react@^19, @types/react-dom@^19, @types/react-syntax-highlighter@^15.5.13, ag-grid-community@^34.2.0, ag-grid-react@^34.2.0, chart.js@^4.5.0, class-variance-authority@^0.7.1, clsx@^2.1.1, eslint@^9, eslint-config-next@15.5.4, lucide-react@^0.544.0, next@15.5.4, next-themes@^0.4.6, react@19.1.0, react-chartjs-2@^5.3.0, react-dom@19.1.0, react-markdown@^10.1.0, react-resizable-panels@^3.0.6, react-rnd@^10.5.2, react-syntax-highlighter@^16.1.0, react-zoom-pan-pinch@^3.7.0, remark-gfm@^4.0.1, tailwind-merge@^3.3.1, tailwindcss@^4, tailwindcss-animate@^1.0.7, tw-animate-css@^1.4.0, typescript@^5 ### Recent commits (newest first) - doc: dashboard redesign plan - feat: dark mode - fix: cors origin - feat: dynamic api url - fix: resolve build errors (eslint, types, hooks) - fix: disable Turbopack for production builds - feat: collapsible code block - fix: distinguish between real code and example code - feat: bring front/back in dashabord, rectangle box - feat: 3 initial suggestion - fix: streaming scrolling mechanism - feat: chat streaming - feat: tabs for data table - feat: collapasible sections - update: clean aiAgent - refactor: main.py and remove undo and quality feature - Delete FETCH_AI_SETUP_GUIDE.md - feat: use groq and data quality rating - update - update ## Key source files (fetched from GitHub, selected and truncated for size) ### docs/architecture-discussion.md ```markdown # Jade AI - Architecture Discussion & Improvement Plan > **Date**: January 14, 2026 > **Status**: In Progress --- ## Overview This document captures ongoing discussions about improving Jade AI's architecture, focusing on transforming it into a more robust BI-style analytics platform similar to Power BI or Tableau. --- ## 1. Dashboard Redesign ### Current State - Free-form canvas with drag/resize (react-rnd) - Chart.js for visualizations - Static charts generated from pandas code - Text and box elements for annotations ### Target State (BI-Style) - Grid-based tile layout with snap-to-grid - Interactive charts with cross-filtering - Global slicers/filters affecting all visuals - Drill-down capabilities - Rich tooltips with additional metrics - Save/load dashboard configurations ### Proposed Stack Changes | Component | Current | Proposed | | ------------- | -------------------------- | ---------------------------------- | | Layout | react-rnd (free-form) | react-grid-layout (grid tiles) | | Charts | Chart.js / react-chartjs-2 | ECharts / echarts-for-react | | State | Local component state | Centralized filter state (Zustand) | | Interactivity | None | Cross-filtering, drill-down | ### Key Features to Implement - [ ] Cross-filtering between charts - [ ] Global slicer/filter components - [ ] Grid-based dashboard layout - [ ] Drill-down navigation - [ ] Rich interactive tooltips --- ## 2. Multi-User & Persistence (Scalable Architecture) ### Current Problems - `df_state` is a Python singleton — all users share the same dataframe - Everything is in-memory — restart = data loss - No user authentication or session isolation ### Target Architecture #### Database Layer | Component | Technology | Purpose | | -------------------- | -------------------- | ---------------------------------------------- | | **Primary DB** | PostgreSQL | Users, sessions, metadata, dashboard configs | | **Cache** | Redis | Session state, hot data, pub/sub for real-time | | **Object Storage** | S3 / MinIO | Uploaded files, exported Parquet files | | **Analytics Engine** | DuckDB (per-request) | Fast analytical queries on Parquet | #### Data Model (PostgreSQL) ```sql -- Users & Auth users (id, email, password_hash, created_at) sessions (id, user_id, created_at, expires_at) -- Projects & Files projects (id, user_id, name, created_at) datasets (id, project_id, name, storage_path, row_count, columns_json, created_at) dataset_versions (id, dataset_id, version, storage_path, created_at, pandas_code) -- Dashboards dashboards (id, project_id, name, layout_json, created_at) dashboard_widgets (id, dashboard_id, type, config_json, position_json) dashboard_filters (id, dashboard_id, column, filter_type, default_valu [truncated — 4308 more characters] ``` ### backend/requirements.txt ``` annotated-types==0.7.0 anyio==4.11.0 cachetools==6.2.0 certifi==2025.10.5 charset-normalizer==3.4.3 click==8.3.0 et_xmlfile==2.0.0 fastapi==0.118.0 groq==0.33.0 h11==0.14.0 httpcore==1.0.7 httpx==0.27.2 idna==3.10 numpy==2.3.3 openpyxl==3.1.5 pandas==2.3.3 pyasn1==0.6.1 pyasn1_modules==0.4.2 pydantic==2.11.9 pydantic_core==2.33.2 python-dateutil==2.9.0.post0 python-dotenv==1.0.0 python-multipart==0.0.20 pytz==2025.2 requests==2.32.5 rsa==4.9.1 six==1.17.0 sniffio==1.3.1 starlette==0.48.0 tenacity==9.1.2 typing-inspection==0.4.2 typing_extensions==4.15.0 tzdata==2025.2 uagents==0.22.10 urllib3==2.5.0 uvicorn==0.37.0 websockets==15.0.1 langgraph==0.2.45 langchain-core==0.3.21 langchain-groq==0.2.1 ``` ### frontend/package.json ``` { "name": "frontend", "version": "0.1.0", "private": true, "scripts": { "dev": "next dev --turbopack", "build": "next build", "start": "next start", "lint": "eslint" }, "dependencies": { "@radix-ui/react-alert-dialog": "^1.1.15", "@radix-ui/react-dialog": "^1.1.15", "@radix-ui/react-dropdown-menu": "^2.1.16", "@radix-ui/react-slot": "^1.2.3", "ag-grid-community": "^34.2.0", "ag-grid-react": "^34.2.0", "chart.js": "^4.5.0", "class-variance-authority": "^0.7.1", "clsx": "^2.1.1", "lucide-react": "^0.544.0", "next": "15.5.4", "next-themes": "^0.4.6", "react": "19.1.0", "react-chartjs-2": "^5.3.0", "react-dom": "19.1.0", "react-markdown": "^10.1.0", "react-resizable-panels": "^3.0.6", "react-rnd": "^10.5.2", "react-syntax-highlighter": "^16.1.0", "react-zoom-pan-pinch": "^3.7.0", "remark-gfm": "^4.0.1", "tailwind-merge": "^3.3.1", "tailwindcss-animate": "^1.0.7" }, "devDependencies": { "@eslint/eslintrc": "^3", "@tailwindcss/postcss": "^4", "@types/node": "^20", "@types/react": "^19", "@types/react-dom": "^19", "@types/react-syntax-highlighter": "^15.5.13", "eslint": "^9", "eslint-config-next": "15.5.4", "tailwindcss": "^4", "tw-animate-css": "^1.4.0", "typescript": "^5" } } ``` ### backend/main.py ```python """Jade AI Backend - Main FastAPI Application""" from dotenv import load_dotenv # Load environment variables FIRST before any other imports load_dotenv() from fastapi import FastAPI from fastapi.middleware.cors import CORSMiddleware from core.config import settings from api.routes import upload_routes, chat_routes, data_routes # Create FastAPI app app = FastAPI(title=settings.APP_TITLE, version=settings.APP_VERSION) # Configure CORS app.add_middleware( CORSMiddleware, allow_origins=settings.CORS_ORIGINS, allow_credentials=True, allow_methods=["*"], allow_headers=["*"], ) # Health check endpoints @app.get("/") def root(): """Root endpoint""" return {"message": "Welcome to Jade AI", "version": settings.APP_VERSION} @app.get("/health") def health_check(): """Health check endpoint""" return {"status": "healthy"} # Include routers app.include_router(upload_routes.router) app.include_router(chat_routes.router) app.include_router(data_routes.router) ``` ### frontend/src/app/layout.tsx ```typescript import type { Metadata } from "next"; import { Geist, Geist_Mono } from "next/font/google"; import "./globals.css"; import { ThemeProvider } from "@/components/theme-provider"; const geistSans = Geist({ variable: "--font-geist-sans", subsets: ["latin"], }); const geistMono = Geist_Mono({ variable: "--font-geist-mono", subsets: ["latin"], }); export const metadata: Metadata = { title: "Jade AI", description: "AI-powered data analysis platform", }; export default function RootLayout({ children, }: Readonly<{ children: React.ReactNode; }>) { return ( {children} ); } ``` ### frontend/src/app/page.tsx ```typescript "use client"; import { useState, useRef, useEffect, useCallback } from "react"; import Image from "next/image"; import { UploadedData, ChatMessage, ChatRequest, ChartData, TextElement, BoxElement } from "../../../shared/types"; import DataTable from "@/components/DataTable"; import Dashboard from "@/components/Dashboard"; import ChatAgent from "@/components/ChatAgent"; import FileExplorer from "@/components/FileExplorer"; import SimpleTabs, { SimpleTab } from "@/components/SimpleTabs"; import { Button } from "@/components/ui/button"; import { DropdownMenu, DropdownMenuContent, DropdownMenuItem, DropdownMenuTrigger, DropdownMenuSeparator, } from "@/components/ui/dropdown-menu"; import { PanelLeftClose, PanelRightClose, LayoutGrid } from "lucide-react"; import { ThemeToggle } from "@/components/ThemeToggle"; import { ResizableHandle, ResizablePanel, ResizablePanelGroup, ImperativePanelHandle, } from "@/components/ui/resizable"; export default function Home() { const [files, setFiles] = useState([]); const [selectedFileIndex, setSelectedFileIndex] = useState(null); const [chatMessages, setChatMessages] = useState([]); const [isLoading, setIsLoading] = useState(false); const [charts, setCharts] = useState([]); const [textElements, setTextElements] = useState([]); const [boxElements, setBoxElements] = useState([]); // Tab system state const [tabs, setTabs] = useState([]); const [activeTabId, setActiveTabId] = useState(null); const [shouldOpenTabs, setShouldOpenTabs] = useState(null); // Refs for collapsible panels const fileExplorerPanelRef = useRef(null); const chatPanelRef = useRef(null); // Track panel collapsed states const [isFileExplorerCollapsed, setIsFileExplorerCollapsed] = useState(false); const [isChatCollapsed, setIsChatCollapsed] = useState(false); // Get the currently selected file const uploadedData = selectedFileIndex !== null ? files[selectedFileIndex] : null; // Tab management functions const openTabForFile = useCallback((fileIndex: number) => { const file = files[fileIndex]; const tabId = `data-${fileIndex}`; // Check if tab already exists const existingTab = tabs.find(t => t.id === tabId); if (existingTab) { setActiveTabId(tabId); return; } // Create new tab const newTab: SimpleTab = { id: tabId, title: file.filename, type: "data", fileIndex: fileIndex, content: null, // Will be rendered dynamically }; setTabs(prev => [...prev, newTab]); setActiveTabId(tabId); }, [files, tabs]); const openDashboardTab = useCallback(() => { const tabId = "dashboard"; // Check if tab already exists const existingTab = tabs.find(t => t.id === tabId); if (existingTab) { setActiveTabId(tabId); return; } // Create new tab const newTab: SimpleTab = { id: tabId, title: "Dashboard", type: "dashboard", content: null, // Will be rendered dynamically }; setTabs(prev => [...prev, newTab]); setActiveTabId(tabId); }, [tabs]); const handleTabClose = (tabId: string) => { const newTabs = tabs.filter(t => t.id !== tabId); setTabs(newTabs); // If closing active tab, switch to last tab if (activeTabId === tabId) { setActiveTabId(newTabs.length > 0 ? newTabs[newTabs.length - 1].id : null); } }; const handleTabReorder = (fromIndex: number, toIndex: number) => { const newTabs = [...tabs]; const [movedTab] = newTabs.splice(fromIndex, 1); newTabs.splice(toIndex, 0, movedTab); setTabs(newTabs); }; const handleTabChange = (tabId: string) => { setActiveTabId(tabId); // If it's a data tab, sync the selected file index if (tabId.startsWith('data-')) { const fileIndex = parseInt(tabId.replace('data-', '')); if (!isNaN(fileIndex) && fileIndex >= 0 && fileIndex < files.length) { setSelectedFileIndex(fileIndex); } } }; // Clear backend state on page load/refresh useEffect(() => { const clearBackendState = async () => { try { const apiUrl = process.env.NEXT_PUBLIC_API_URL || 'http://localhost:8000'; await fetch(`${apiUrl}/clear`, { method: 'POST', }); } catch (error) { console.error('Error clearing backend state:', error); } }; clearBackendState(); }, []); // Open tabs after file upload useEffect(() => { if (shouldOpenTabs !== null && files.length > shouldOpenTabs) { const dataTabId = `data-${shouldOpenTabs}`; // Open data tab first (this will be the active tab) openTabForFile(shouldOpenTabs); // Open dashboard tab but don't focus it const dashboardTabId = "dashboard"; const existingDashboardTab = tabs.find(t => t.id === dashboardTabId); if (!existingDashboardTab) { const newTab: SimpleTab = { id: dashboardTabId, title: "Dashboard", type: "dashboard", content: null, }; setTabs(prev => [...prev, newTab]); // Keep focus on data tab by setting it as active setActiveTabId(dataTabId); } setShouldOpenTabs(null); } }, [files, shouldOpenTabs, tabs, openTabForFile]); // Keyboard shortcuts for toggling panels useEffect(() => { const handleKeyDown = (e: KeyboardEvent) => { // Check if Cmd (Mac) or Ctrl (Windows/Linux) is pressed const isCmdOrCtrl = e.metaKey || e.ctrlKey; if (!isCmdOrCtrl) return; switch (e.key) { case '1': // Cmd/Ctrl + 1: Toggle Files e.preventDefault(); if (isFileExplorerCollapsed) { fileExplorerPanelRef.current?.expand(); } else { fileExplorerPanelRef.current?.collapse(); [truncated — 18897 more characters] ``` ### frontend/next.config.ts ```typescript import type { NextConfig } from "next"; const nextConfig: NextConfig = { /* config options here */ }; export default nextConfig; ``` ### shared/types.ts ```typescript export interface UploadedData { filename: string; rows: number; columns: number; column_names: string[]; dtypes: Record; preview: Record[]; data: Record[]; sheet_name?: string; // Optional: which sheet this data represents original_filename?: string; // Optional: the original file name before adding sheet suffix } export interface ChartData { type: | "line" | "bar" | "pie" | "doughnut" | "area" | "bubble" | "polarArea" | "radar" | "scatter" | "mixed"; labels: string[]; datasets: { label: string; data: | number[] | { x: number; y: number }[] | { x: number; y: number; r: number }[]; borderColor?: string | string[]; backgroundColor?: string | string[]; fill?: boolean; tension?: number; type?: "line" | "bar" | "scatter"; pointRadius?: number; pointHoverRadius?: number; }[]; title: string; } export interface ChatMessage { role: "user" | "assistant"; content: string; pandas_code?: string; has_code?: boolean; data_updated?: boolean; error?: string; print_output?: string; narrative_output?: string; chart_data?: ChartData; } export interface ChatRequest { message: string; chat_history: ChatMessage[]; } export interface TextElement { id: string; content: string; x: number; y: number; width: number; height: number; fontSize?: number; fontWeight?: string; color?: string; } export interface BoxElement { id: string; x: number; y: number; width: number; height: number; backgroundColor?: string; borderColor?: string; borderWidth?: number; borderRadius?: number; } export interface ChatResponse { response: string; pandas_code: string | null; has_code: boolean; data_updated: boolean; updated_data: { data: Record[]; columns: string[]; rows: number; dtypes: Record; } | null; error: string | null; print_output: string | null; narrative_output: string | null; chart_data: ChartData | null; } ``` ### backend/api/__init__.py ```python """API layer for Jade AI backend""" from . import models from . import routes __all__ = ["models", "routes"] ``` ### backend/workflows/__init__.py ```python """LangGraph workflows for Jade AI""" from .orchestrator import create_workflow_orchestrator __all__ = ["create_workflow_orchestrator"] ``` [44 more indexed source files omitted to keep this export small. The full file list is in the Codebase structure section above.]