Devika AI

What is Devika AI?

Devika AI is an agentic AI software engineer designed to understand high-level human instructions, break them down into steps, research relevant information, and write code to achieve a given objective. It aims to be a competitive open-source alternative to Devin by Cognition AI.

What are the key features of Devika AI?

Devika AI boasts several key features, including:

• Support for various large language models (Claude 3, GPT-4, Gemini, Mistral, Groq, and local LLMs via Ollama)
• Advanced AI planning and reasoning
• Contextual keyword extraction for research
• Seamless web browsing
• Code writing in multiple languages
• Dynamic agent state tracking
• Natural language interaction
• Project-based organization

Is Devika AI an open-source project?

Yes, Devika AI is an open-source project. Its code is publicly available on GitHub, and contributions are welcome to help enhance its capabilities and performance.

How can I install and run Devika AI?

To install Devika AI, you need Python (3.10 to 3.12), Node.js (>=18), and Bun. After cloning the repository, you'll create a virtual environment, install dependencies using uv pip install -r requirements.txt, install Playwright browsers, and then start the Devika server with python devika.py. For the frontend, navigate to the ui/ directory, run bun install, and then bun run start.

Does Devika AI have an official website?

No, Devika AI does not have an official website. The project is currently in an early development/experimental stage, and its primary presence is on GitHub.

What kind of tasks can Devika AI perform?

Devika AI can assist with complex coding tasks, such as creating new features, fixing bugs, or developing entire projects from scratch, all with minimal human guidance. It processes high-level objectives, breaks them down, and generates code.

What API keys are required for Devika AI to function?

Devika AI requires API keys for various services, including Bing, Google Search, OpenAI, Gemini, Claude, Mistral, Groq, and Netlify. These can be configured via the settings page in the UI after the initial setup.

Where can I find support or contribute to Devika AI?

For support, you can raise an issue in the GitHub issue tracker or join the discussions. Contributions are welcome, and you can find guidelines in the CONTRIBUTING.md file. There is also a Discord server for the Devika community.

What is the license for Devika AI?

Devika AI is released under the MIT License.

Devika AI GitHub Link

https://github.com/stitionai/devika

Devika AI: An Open-Source Agentic AI Software Engineer

Overview

Devika AI is an open-source, agentic AI software engineer. It's designed to understand complex human instructions, break them down into actionable steps, conduct research, and write code to achieve specified objectives. Devika aims to be a competitive open-source alternative to proprietary AI software engineers like Devin by Cognition AI, with an ambitious goal to match and eventually surpass its performance in benchmarks like SWE-bench.

Main Purpose and Target User Group

• Main Purpose: To revolutionize software development by providing an AI pair programmer capable of handling complex coding tasks with minimal human guidance, including creating new features, fixing bugs, or developing entire projects.
• Target User Group: Developers, software engineers, and teams looking for an advanced AI assistant to automate and streamline their coding workflows, particularly those interested in open-source solutions and leveraging large language models for software engineering tasks.

Function Details and Operations

• Instruction Understanding: Interprets high-level human instructions for software development tasks.
• Task Decomposition: Breaks down complex objectives into manageable, sequential steps.
• Information Research: Utilizes web browsing capabilities and contextual keyword extraction to gather relevant information.
• Code Generation: Writes code in multiple programming languages based on research and task requirements.
• LLM Support: Supports various large language models including Claude 3, GPT-4, Gemini, Mistral, Groq, and local LLMs via Ollama. Claude 3 models are recommended for optimal performance.
• Dynamic State Tracking: Visualizes and tracks the agent's progress and current state.
• Chat Interface: Enables natural language interaction for providing instructions and receiving updates.
• Project Management: Organizes and manages software development tasks on a project-by-project basis.
• Extensible Architecture: Designed for easy integration of new features and functionalities.
• Web Interface: Accessible via a web browser for user interaction and project management.
• Configuration: Allows users to configure API keys for various services (Bing, Google Search, OpenAI, Gemini, Claude, Mistral, Groq, Netlify) and API endpoints (Bing, Google, Ollama, OpenAI).

User Benefits

• Increased Efficiency: Automates significant portions of the software development process, accelerating project completion.
• Reduced Manual Effort: Minimizes the need for human intervention in coding and research tasks.
• Enhanced Productivity: Acts as an AI pair programmer, allowing developers to focus on higher-level design and problem-solving.
• Access to Advanced AI: Leverages state-of-the-art LLMs and AI planning capabilities for sophisticated code generation.
• Open-Source Flexibility: Provides a transparent and customizable alternative to proprietary solutions.
• Streamlined Workflow: Integrates research, planning, and coding into a single, cohesive process.

Compatibility and Integration

• Operating System: Requires Python >= 3.10 and < 3.12, NodeJs >= 18, and Bun.
• Browser Integration: Uses Playwright for browsing capabilities.
• API Integrations: Configurable with API keys for various search engines (Bing, Google Search) and LLM providers (OpenAI, Gemini, Anthropic, Mistral, Groq).
• Deployment: Supports Netlify API key for deploying and managing web projects.
• Local LLM Support: Integrates with Ollama for running local language models.

Customer Feedback and Case Studies

• Currently in early development/experimental stage with many unimplemented/broken features.
• Aims to be a competitive open-source alternative to Devin by Cognition AI, with an "overly ambitious" goal to meet and eventually beat its SWE-bench Benchmarks score.
• Community contributions are welcomed to accelerate development and feature implementation.
• Demos are available (e.g., devika-pygame-demo.mp4) showcasing its capabilities.

Access and Activation Method

• Source Code: Available on GitHub: https://github.com/stitionai/devika.
• Installation:
  1. Clone the GitHub repository.
  2. Navigate to the project directory.
  3. Create a virtual environment and install dependencies using uv venv and uv pip install -r requirements.txt.
  4. Install Playwright for browser capabilities: playwright install --with-deps.
  5. Start the Devika server: python devika.py.
  6. For the frontend, navigate to the ui/ directory, run bun install, and then bun run start.
• Access: Access the Devika web interface by navigating to http://127.0.0.1:3001 in a web browser.
• Configuration: API keys and endpoints are configured via the settings page in the UI, which creates a config.toml file in the root directory.