Codel is a fully autonomous AI agent designed to execute complex tasks and projects through a terminal, browser, and editor.
Purpose: Codel is used to automate development or system tasks that require multi-step operations, and can autonomously determine and complete the next action.
Main capabilities:
- Fully autonomous: can automatically detect the next step and execute it.
- Secure: all operations run in a sandboxed Docker environment.
- Built-in browser: can obtain the latest information from the web (such as tutorials, documentation, etc.).
- Built-in text editor: can view all modified files in the browser.
- History: all commands and outputs are saved in a PostgreSQL database.
- Automatic Docker image selection: automatically matches the appropriate Docker image based on the user's task.
How to use: The simplest way to run it is to use a prebuilt Docker image. The latest image can be found on the GitHub Container Registry. When using any supported language model, the corresponding environment variables need to be set. The Docker image can be run with the following command, deleting or changing environment variables as needed: (Note: the specific command text is not provided in the materials) Or create a .env file and run the Docker image with the --env-file parameter; for more information, refer to the relevant documentation. After that, visit localhost:3000 in the browser to start using Codel. Development information can be found in DEVELOPMENT.md, and the project roadmap can be found here.
Applicable scenarios: suitable for scenarios that require automated handling of complex tasks and projects, such as development, debugging, information gathering, etc.
Limitations: no other limitation information is provided in the materials.
