codel - Autonomous AI Agent in Secure Docker for Advanced Task Execution

Introduction to Codel

Codel represents a cutting-edge innovation in artificial intelligence, offering a fully autonomous AI agent capable of tackling complex tasks and projects. This AI operates with remarkable efficiency using tools such as terminals, browsers, and text editors, streamlining processes that typically require human intervention.

Key Features

Codel distinguishes itself with several noteworthy features designed to enhance user experience and efficiency:

Secure Environment: Codel ensures a secure operation by running exclusively within a sandboxed Docker environment. This isolated setting minimizes security risks by preventing external interference.
Autonomy: The AI's inherent autonomy allows it to independently determine the next steps in any given task, executing them seamlessly without requiring manual guidance.
Built-In Browser: Information retrieval is one of Codel's strengths. It features an integrated browser that fetches the latest tutorials and documentation from the web when necessary, keeping the workflow current and well-informed.
Text Editor: Modifications to files can be easily viewed in the browser itself thanks to the built-in text editor, ensuring transparency and ease of access.
History Logging: Every command and output generated during the operation is documented and securely stored in a PostgreSQL database, allowing for an exhaustive audit trail.
Docker Image Picker: To streamline operations, Codel automatically selects the appropriate Docker image based on the task at hand, ensuring optimal performance.
Self-Hosted and Modern UI: Users have the freedom to self-host their instances with a sleek, contemporary user interface that enhances interaction.

Getting Started with Codel

For those eager to deploy Codel, the simplest method is utilizing a pre-built Docker image. This image can be accessed from the GitHub Container Registry. Proper configuration is essential, involving setting specific environment variables to match the language models being used.

Here's a snippet of how to run the Docker image:

docker run \
  -e OPEN_AI_KEY=your_open_ai_key \
  -e OPEN_AI_MODEL=gpt-4-0125-preview \
  -e OLLAMA_MODEL=llama2 \
  -p 3000:8080 \
  -v /var/run/docker.sock:/var/run/docker.sock \
  ghcr.io/semanser/codel:latest

Alternatively, a .env file can be employed alongside the --env-file flag for configuration, with additional details available here. Once Codel is running, users can visit localhost:3000 to start utilizing its features.

Development and Roadmap

For those interested in contributing or understanding more about Codel’s technical underpinnings, the comprehensive development guide can be found in the DEVELOPMENT.md. Future plans and the project's trajectory are outlined in the project's roadmap.

Acknowledgements

Codel's development has been influenced and supported by various projects and resources, including insightful arXiv papers and invaluable tools like go-rod and JsonGenius. These contributions have played a vital role in Codel’s inception and evolution.