Devika

Autonomous Agents 💻 Python ⚖️ MIT 🔴 No recent commits
19.6k stars

Devika is an open-source AI software engineer that interprets high-level instructions, plans, researches, and writes code. It aims to automate complex coding tasks with minimal human guidance, serving as an alternative to Devin.

Devika demo
🖼️ Screenshot from the project README

✨ Key features

  • Supports Claude 3, GPT-4, Gemini, Mistral, Groq, and local LLMs via Ollama.
  • Advanced AI planning and reasoning for task decomposition.
  • Contextual keyword extraction for focused web research.
  • Seamless web browsing and information gathering.
  • Writes code in multiple programming languages.
  • Dynamic agent state tracking and visualization.

🎯 Use cases

  • Automate feature development from natural language descriptions.
  • Fix bugs by providing a high-level problem statement.
  • Generate entire project scaffolds from scratch.
  • Research and integrate external APIs or libraries.
  • Assist in code reviews and refactoring tasks.

📦 Installation

🧰 Requirements: Requires Python >=3.10 and <3.12, Node.js >=18, bun, and optionally API keys for cloud LLMs or Ollama for local models.

git clone https://github.com/stitionai/devika.git
cd devika
uv venv
On macOS and Linux.
source .venv/bin/activate
On Windows.
.venv\Scripts\activate
uv pip install -r requirements.txt
playwright install --with-deps
python devika.py

Then in a new terminal:

cd ui/
bun install
bun run start

Access the web interface at http://127.0.0.1:3001.

🚀 Usage

  1. Open the web interface.
  2. Create a new project via 'select project' > 'new project'.
  3. Choose search engine and model configuration.
  4. In chat, provide a high-level objective.
  5. Monitor progress and review generated code.
  6. Iterate with further instructions.

⚠️ Good to know

This project is in very early development/experimental stage with many unimplemented or broken features.

❓ FAQ

What models does Devika support?

Devika supports Claude 3, GPT-4, Gemini, Mistral, Groq, and local LLMs via Ollama. For optimal performance, use Claude 3 models.

How do I configure API keys?

On first run, Devika creates a config.toml file. You can set API keys for services like Bing, Google, OpenAI, Gemini, Claude, Mistral, Groq, and Netlify via the settings page in the UI.

Is Devika production-ready?

No, it is in a very early development/experimental stage with many unimplemented or broken features. Contributions are welcome.

Can I use local models without API keys?

Yes, you can use local LLMs via Ollama. Follow the Ollama setup guide in the docs to configure it.

📊 Repository

Stars★ 19,558
Forks🍴 2,580
Open issues🐛 196
Last commit🕒 Sep 25, 2025
Created📅 Mar 2024
Language💻 Python
License⚖️ MIT

🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.