Skip to content

About

An agentic GenAI software development platform that transforms natural-language requirements into complete, tested, reviewed, and documented software projects using multiple AI agents.

Topics

Resources

Contributing

Security policy

Stars

2 stars

Watchers

0 watching

Forks

Latest commit

Β 

History

10 Commits

Folders and files

Repository files navigation

πŸ€– MyAppMyWeb (DevAgent AI)

Autonomous Multi-Agent Software Development & Self-Healing Platform

LangGraph FastAPI Next.js Docker TypeScript Tailwind CSS Security License


A state-of-the-art autonomous software engineering platform that transforms natural language prompts into production-grade, tested, and self-healed full-stack codebases.
Multi-Agent LangGraph DAG β€’ Ephemeral Docker Code Sandboxing β€’ Automated Jest/Supertest/Pytest Harnesses β€’ Real-Time Self-Healing Feedback Loops β€’ Interactive Live App Preview β€’ 1-Click GitHub & ZIP Export

UI Showcase β€’ Key Capabilities β€’ Architecture β€’ Security & Sandboxing β€’ Quick Start β€’ API Reference β€’ Author


πŸ–₯️ UI Command Center Showcase

1. Master Command Center & Multi-Agent LangGraph DAG

Autonomous SWE Prompt Engine with presets, visual 5-Agent DAG pipeline, 9-stage task checklist, and multi-file code viewer.

Master Command Center & Multi-Agent DAG


2. Ephemeral Docker Sandbox Stream & Autonomous Self-Healing Loop

Real-time containerized test execution catching assertion errors and automatically triggering developer agent self-healing loops.

Docker Sandbox Stream & Self-Healing


3. Interactive Live App Preview Sandbox

Instant live application sandbox with responsive device switcher (Desktop, Tablet, Mobile), theme toggle, and interactive state.

Interactive Live App Preview


4. 1-Click Export & Direct Push to GitHub Integration

Publish generated full-stack repositories directly to GitHub with automated manifests or download complete ZIP bundles.

Direct GitHub Export Modal


🌟 Key Capabilities & Workflow

1. 🧠 Multi-Agent Orchestration (LangGraph State Machine)

Unlike naive single-prompt code generators, MyAppMyWeb orchestrates 5 specialized autonomous agents connected in a stateful execution graph:

β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”     β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  πŸ‘‘ Manager     β”‚ ──► β”‚  πŸ’» Developer    β”‚ ──► β”‚  πŸ§ͺ Tester      β”‚ ──► β”‚  πŸ›‘οΈ Reviewer     β”‚ ──► β”‚  πŸ“ Doc Agent    β”‚
β”‚  Spec & Task DAGβ”‚     β”‚  Code Synthesis  β”‚     β”‚  Docker Sandbox β”‚     β”‚  Security & AST  β”‚     β”‚  README & OpenAPIβ”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜     β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               β–²                          β”‚
                               └────── Failing Trace β”€β”€β”€β”€β”€β”˜
                                    (Self-Healing Loop)
Agent Responsibility Core Deliverables
πŸ‘‘ Manager Agent Natural Language Spec Decomposition Technical Architecture, Tech Stack Selection, and 9-Stage Task Execution DAG
πŸ’» Developer Agent Full-Stack Multi-File Synthesis 15+ Source Files across Frontend (React/Next.js) & Backend (Express/FastAPI/MongoDB)
πŸ§ͺ Tester Agent Automated Test Generation & Sandboxing Unit & Integration Test Suites (Jest, Supertest, Pytest) executed in Docker
πŸ”„ Self-Healing Router Failure Trace Inspection & Auto-Correction Real-time AST error diagnosis and Before/After Git Diff patches
πŸ›‘οΈ Reviewer Agent Static Code Analysis & Security Audit Grade A+ (96/100) Quality Scorecard, OWASP Top 10 compliance check
πŸ“ Doc Agent Production-Ready Documentation Comprehensive README, API documentation, and OpenAPI 3.0 specifications

2. πŸ”„ Closed-Loop Self-Healing Engine

When an assertion fails inside the Docker test sandbox, the platform automatically intercepts the error, locates the failing line, and triggers a surgical repair:

flowchart TD
    Tester["πŸ§ͺ Tester Agent: Executes Test Suite in Docker"] --> AssertCheck{"All Tests Pass?"}
    AssertCheck -- Yes --> Reviewer["πŸ›‘οΈ Reviewer Agent: Security & AST Audit"]
    AssertCheck -- No --> TraceExtractor["Extract Failing File, Line Number & AssertionError"]
    TraceExtractor --> DevPatch["πŸ’» Developer Agent: Synthesizes Corrected Code"]
    DevPatch --> DiffRecord["Record Unified Git Diff: Old vs Self-Healed"]
    DiffRecord --> SandboxRetest["🐳 Docker Sandbox: Re-run Test Harness"]
    SandboxRetest --> AssertCheck
    Reviewer --> DocAgent["πŸ“ Doc Agent: OpenAPI & Final Artifacts"]
Loading

3. πŸ–₯️ Interactive Live App Preview Sandbox

  • Responsive Viewport Toggles: Seamlessly test generated apps across πŸ–₯️ Desktop (100%), πŸ’» Tablet (720px), and πŸ“± Mobile (375px).
  • Isolated Iframe Execution: Live interactivity without host machine contamination.
  • Dynamic State Simulation: Add, filter, complete, and delete records in real-time.
  • Theme Support: Integrated dark and light mode toggle.

4. πŸ“¦ 1-Click Export & GitHub Integration

  • Download Standalone ZIP: Instantly bundle the entire generated workspace into a ready-to-run archive.
  • Push to GitHub: Direct Personal Access Token / OAuth integration to scaffold a new remote repository with atomic commit history.

πŸ›‘οΈ Security Architecture & Sandbox Boundaries

MyAppMyWeb executes generated code inside hardened ephemeral environments to prevent Arbitrary Code Execution (ACE) and sandbox escapes:

Control Specification Security Objective
Network Isolation --network none Prevents unauthorized telemetry, external network access, and reverse shells.
User Privileges UID 1000:1000 (Non-Root) Disables root privilege escalation inside the container.
Root Filesystem read-only Protects base system libraries and container binaries from unauthorized tampering.
Linux Capabilities --cap-drop=ALL Drops all raw kernel, socket, and mounting capabilities.
Resource Limits 512MB RAM, 1.0 CPU, 64 PIDs Prevents fork bombs and Host Out-of-Memory (OOM) resource starvation.
Execution Watchdog 30s Strict Timeout Terminates blocking or infinite loops automatically.

Detailed documentation: SECURITY.md β€’ security/threat-model.md β€’ security/sandbox-policy.md


πŸ—οΈ System Architecture

flowchart TD
    subgraph UI["Frontend: Next.js 15 Command Center"]
        PromptBox["Autonomous SWE Prompt Engine"]
        AgentDAG["Multi-Agent State Graph Visualizer"]
        TaskBoard["Task DAG Checklist (9 Stages)"]
        FileTree["Hierarchical Explorer & Syntax Viewer"]
        DiffView["Self-Healing Before/After Diff Viewer"]
        LivePreview["Interactive App Preview Sandbox"]
        Terminal["Live Docker ANSI Log Stream (SSE)"]
    end

    subgraph API["Backend: FastAPI (Python 3.11)"]
        SSEStream["POST /api/v1/agent/run/stream"]
        ExportAPI["GET /api/v1/agent/export-zip"]
        GitHubAPI["POST /api/v1/agent/push-github"]
        HealthAPI["GET /health"]
    end

    subgraph Orchestration["Agent Execution Engine (LangGraph)"]
        Manager["πŸ‘‘ Manager Agent"]
        Developer["πŸ’» Developer Agent"]
        Tester["πŸ§ͺ Tester Agent"]
        Reviewer["πŸ›‘οΈ Reviewer Agent"]
        DocAgent["πŸ“ Documentation Agent"]
    end

    subgraph Sandboxing["Execution Safety"]
        DockerRunner["Docker Ephemeral Sandbox / Subprocess"]
        VirtualFS["In-Memory Virtual AST Filesystem"]
    end

    PromptBox --> SSEStream --> Manager --> Developer --> VirtualFS --> DockerRunner --> Tester
    Tester -. Failing Trace .-> Developer
    Tester -- 100% Passed --> Reviewer --> DocAgent
    VirtualFS --> ExportAPI & GitHubAPI
    DockerRunner -. ANSI Stream .-> Terminal
Loading

πŸ› οΈ Quick Start Guide

Prerequisites

  • Python 3.10+
  • Node.js 18+ & npm
  • (Optional) Docker Desktop (for containerized execution)

1. Start Backend Server

cd backend
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt
python -m uvicorn app.main:app --port 8000 --reload

2. Start Frontend Command Center

cd frontend
npm install
npm run dev

Open http://localhost:3000 in your browser!


πŸ“Š API Reference

Method Endpoint Description
POST /api/v1/agent/run/stream Stream real-time agent execution events, DAG progress, and Docker logs via SSE
GET /api/v1/agent/export-zip Download the generated project files as a compressed ZIP archive
POST /api/v1/agent/push-github Export the generated codebase directly to a remote GitHub repository
GET /health Service health check and uptime status

πŸ“‚ Repository Structure

MyAppMyWeb/
β”œβ”€β”€ .github/
β”‚   β”œβ”€β”€ workflows/
β”‚   β”‚   β”œβ”€β”€ ci.yml                # CI Test & Verification Pipeline
β”‚   β”‚   └── security-scan.yml     # CodeQL & Security Scanner
β”‚   └── dependabot.yml            # Automated Dependency Management
β”œβ”€β”€ docs/
β”‚   └── images/                   # High-Resolution UI Screenshots
β”‚       β”œβ”€β”€ dashboard-overview.png
β”‚       β”œβ”€β”€ docker-stream-self-healing.png
β”‚       β”œβ”€β”€ live-app-preview.png
β”‚       └── github-export-modal.png
β”œβ”€β”€ backend/
β”‚   β”œβ”€β”€ app/
β”‚   β”‚   β”œβ”€β”€ agents/               # Multi-Agent Definitions
β”‚   β”‚   β”‚   β”œβ”€β”€ manager.py        # Spec & Task Decomposition
β”‚   β”‚   β”‚   β”œβ”€β”€ developer.py      # Code Generation & Self-Healing
β”‚   β”‚   β”‚   β”œβ”€β”€ tester.py         # Test Harness Runner
β”‚   β”‚   β”‚   β”œβ”€β”€ reviewer.py       # Static Code & AST Security Audit
β”‚   β”‚   β”‚   └── documentation.py  # Markdown & OpenAPI Generator
β”‚   β”‚   β”œβ”€β”€ api/                  # FastAPI REST & SSE Endpoints
β”‚   β”‚   β”œβ”€β”€ graph/                # LangGraph StateGraph Engine
β”‚   β”‚   β”œβ”€β”€ sandbox/              # Docker & Virtual FS Sandboxing
β”‚   β”‚   └── main.py               # Application Entrypoint
β”‚   β”œβ”€β”€ Dockerfile
β”‚   └── requirements.txt
β”œβ”€β”€ frontend/
β”‚   β”œβ”€β”€ src/
β”‚   β”‚   β”œβ”€β”€ app/                  # Next.js 15 App Router
β”‚   β”‚   └── components/           # Command Center UI Components
β”‚   β”‚       β”œβ”€β”€ Header.tsx        # Telemetry & Status Bar
β”‚   β”‚       β”œβ”€β”€ AgentDAGVisualizer.tsx
β”‚   β”‚       β”œβ”€β”€ TaskBoard.tsx     # 9-Stage Task Execution Checklist
β”‚   β”‚       β”œβ”€β”€ CodeEditorView.tsx# Explorer & Code Viewer
β”‚   β”‚       β”œβ”€β”€ DiffViewer.tsx    # Self-Healing Git Diff Comparison
β”‚   β”‚       β”œβ”€β”€ LiveAppPreview.tsx# Interactive App Simulator
β”‚   β”‚       β”œβ”€β”€ ScorecardReview.tsx # Security & Quality Scorecard
β”‚   β”‚       β”œβ”€β”€ TerminalLogStream.tsx
β”‚   β”‚       └── GitHubModal.tsx   # Direct Push Dialog
β”‚   β”œβ”€β”€ Dockerfile
β”‚   └── package.json
β”œβ”€β”€ security/
β”‚   β”œβ”€β”€ threat-model.md           # STRIDE & OWASP Mitigation Matrix
β”‚   └── sandbox-policy.md         # Hardening & Security Standards
β”œβ”€β”€ docker-compose.yml
β”œβ”€β”€ SECURITY.md
β”œβ”€β”€ CONTRIBUTING.md
β”œβ”€β”€ LICENSE
└── README.md

πŸ‘€ Author

Developed with ❀️ by Ishant6565.


πŸ“„ License

This project is licensed under the MIT License β€” see the LICENSE file for details.

About

An agentic GenAI software development platform that transforms natural-language requirements into complete, tested, reviewed, and documented software projects using multiple AI agents.

Topics

Resources

Contributing

Security policy

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages