Helix DevAgent AI is a futuristic, autonomous agentic suite designed to evolve the Software Development Life Cycle (SDLC).
Powered by Google Gemini 2.5 Flash, it serves as a neural interface where developers command specialized AI agents to handle unit testing, debugging, code auditing, and architectural strategy.
Modern software engineering involves a high "context-switching tax." Developers spend hours writing boilerplate tests, parsing obscure error logs, and debating style in code reviews. Existing tools are fragmented across CLIs, CI/CD pipelines, and web portals.
Helix DevAgent AI unifies these workflows into a Single Pane of Glass. Instead of manually writing a Jest suite, you activate the Test Forge. Instead of Googling a stack trace, you deploy the Auto-Debugger. It transforms the IDE from a text editor into an Intelligent Command Center.
- Role: Central Intelligence & Strategy.
- Capabilities: Answers high-level architectural questions, explains complex DevOps concepts, and retains context across the session in a persistent local database.
- UI: High-contrast, sharp interface for maximum readability.
- Role: Codebase Navigation & Analysis.
- Capabilities: Upload local files/folders. Features a floating "Summarize File" button for instant AI summaries of code logic. Allows selecting code snippets to "Explain" or "Implement" new features directly via the Assistant.
- Input: Raw source code (Python, JS, Go, Java, etc.).
- Output: Comprehensive unit test suites (Jest, PyTest, JUnit).
- Behavior: Covers happy paths, edge cases, and mocks dependencies automatically.
- Input: Broken code + Error logs/Stack traces.
- Output: Root cause analysis and fixed code snippets.
- Behavior: Correlates logic errors with runtime exceptions to provide a precise fix.
- Input: Code snippets.
- Output: Detailed Markdown report covering Security (OWASP), Performance, and Clean Code principles.
- Persistence: Automatically saves review history for compliance auditing.
- Input: Unstructured raw server logs.
- Output: Structured incident reports with severity classification and remediation steps.
- Input: Legacy or "Spaghetti" code.
- Output: Modernized, optimized, and readable code reducing cyclomatic complexity.
- Privacy First: All data is stored locally in the browser (LocalStorage).
- Portability: Export your entire session history to a JSON database file (
helix_db.json) and import it on any machine.
The Project Explorer is the neural hub for interacting with your local codebase. It provides a file-system interface directly within the browser, augmented by AI context actions.
- Trigger: Click the "Summarize" button in the toolbar or the floating action button that appears on hover.
- Function: Sends the entire file content to the Helix Assistant.
- Output: A concise, bullet-point executive summary of the file. It highlights:
- Primary Responsibility: What the file does.
- Key Functions: Important methods and their purposes.
- Dependencies: Critical imports and external connections.
- Trigger: Click the "Implement" button.
- Function: Opens a dialogue with the Assistant pre-loaded with the file's context.
- Workflow: You describe a new feature (e.g., "Add a dark mode toggle" or "Handle API error retries"). The AI generates the specific code modifications required to implement that feature within the existing file structure.
- Full File Explanation: The "Explain" button provides a comprehensive walkthrough of the entire file's logic, data flow, and architecture.
- Smart Selection: Highlight any snippet of text within the code viewer. A "Explain Selection" button dynamically appears. Clicking this focuses the AI on just that specific block, perfect for decoding complex algorithms or regex patterns without losing context of the surrounding code.
Helix is designed as a modern React application. While it can run in containerized AI environments, you can also run it locally.
- Node.js (v18 or higher)
- Gemini API Key: Obtainable from Google AI Studio.
-
Clone the Repository
git clone https://github.com/ksharma6309/helix-devagent.git cd helix-devagent -
Install Dependencies (If using a bundler like Vite/Create React App)
npm install
-
Configure Environment Create a
.envfile in the root directory:REACT_APP_API_KEY=your_actual_gemini_api_key
Note: The application expects
process.env.API_KEYto be available. -
Run the Application
npm start # or npm run dev -
Access Helix Open your browser to
http://localhost:3000(or the port specified by your runner).
We chose Google AI Studio and the Gemini 2.5 Flash model as the core engine for Helix because of its specific advantages for real-time developer tools.
- Low Latency & High Throughput: Gemini 2.5 Flash is optimized for speed, making real-time code generation and debugging feel instantaneous.
- Large Context Window: It can ingest massive log files or long code snippets (up to 1M tokens) without losing context.
- Multimodal Capabilities: It can natively understand text, code, and structured data seamlessly.
- Frontend Framework: React 19
- Language: TypeScript
- Styling: Tailwind CSS (Cyber-Glass Aesthetic with High Contrast)
- AI Engine: Google Gemini API (
gemini-2.5-flash) - Icons: Lucide React
- Charts: Recharts
- Build Tooling: ES Modules (No-bundler setup for rapid prototyping)
/
βββ index.html # Entry point
βββ src/
β βββ App.tsx # Main Router & Layout
β βββ components/ # Reusable UI (Buttons, Sidebar, Markdown)
β βββ services/ # AI Integration (Gemini Service)
β βββ utils/ # Storage & Persistence Logic
β βββ views/ # Main Application Screens
β β βββ Dashboard.tsx # Command Center
β β βββ ChatAssistant.tsx # Helix Assistant
β β βββ ProjectExplorer.tsx # File Navigation & Summarization
β β βββ TestGenerator.tsx # Test Forge
β β βββ Debugger.tsx # Auto-Debugger
β β βββ CodeReview.tsx # Code Auditor
β β βββ DatabaseManager.tsx # Data Vault
β βββ types.ts # TypeScript Interfaces
βββ README.mdThis project is licensed under the MIT License - see the LICENSE file for details.
Khushboo Sharma
GitHub: https://github.com/ksharma6309
LinkedIn: https://www.linkedin.com/in/khushboo-sharma-b5b372125/






