The DOXA CLI is the primary tool for uploading and submitting agents to DOXA: a powerful platform for hosting engaging artificial intelligence competitions.
🔗 Compete on DOXA
➡️ Join the DOXA Community on Discord
The DOXA CLI may be installed using pip
in the following way:
pip install doxa-cli
Assuming you have your Python Scripts
folder on your PATH
, you can use the DOXA CLI via the doxa
command in your shell. Otherwise, you will have to use python -m doxa_cli
(or python3 -m doxa_cli
on some systems) to access DOXA CLI commands.
Before you can upload agents to DOXA using the CLI, you must first log in with your DOXA account by running the following command:
doxa login
If supported, this will open an authorisation prompt in your default browser. If you are not already logged in on doxaai.com, you will first be invited to do so. Thereafter, you wil be asked to authorise the DOXA CLI to have access to your DOXA account. Once you signify your approval, the DOXA CLI will automatically log you in within a few seconds.
By default, the DOXA CLI will store its configuration in a location that follows the norm for your operating system (e.g. within %APPDATA%\doxa\doxa
on Windows, ~/Library/Application Support/doxa
on macOS and ~/.config/doxa
on Linux). You can find this location by running doxa config
. If for whatever reason you would like to store the DOXA CLI configuration elsewhere (e.g. due to a permissions issue), you may use a different directory by setting the DOXA_CONFIG_DIRECTORY
environment variable.
You may log out by running the following command:
doxa logout
You can retrieve information on the user account with which you are currently logged in by performing the following command:
doxa user
You may upload and submit agents to DOXA using the following command:
doxa upload [AGENT DIRECTORY]
The agent directory must contain a doxa.yaml
file informing DOXA how to process your submission.
The following is an example of the contents of one such doxa.yaml
file:
competition: uttt
environment: cpu
language: python
entrypoint: evaluate.py