Every algorithm on Algorithmia has a dedicated git repository. Using this repository allows you to seamlessly integrate Algorithmia with your current workflows, such as your code editor, IDE, or deployment pipelines. If you’re unfamiliar with git we recommend this brief guide.
Accessing a repository
You can clone existing algorithms from the following endpoint:
:algoname parts with the correct values, case sensitive. If you are part of an organization,
:username with the name of your org.
Committing changes to an algorithm
First pull from origin to make sure you have the latest changes to the master branch.
Then add your changes as usual.
Once ready, push to Algorithmia.
Everything will behave as a normal git repository, with the additional post-commit hook that will compile your master branch, create a private REST API endpoint, and print out that API endpoint to your console.
The output of git push will print a last line that looks something like this:
This endpoint is private to you only (and your organization if you are part of one), and behaves just like a public/published algorithm except for the fact that it is not listed in the marketplace and will only accept calls from your API keys. This is a great way to run regression tests against an algorithm before making it available to others.
Testing your changes
With the private REST API endpoint in hand, you can run your unit tests against it or manually test that algorithm from the command line using the Algorithmia CLI. Testing using the CLI will be something similar to:
See the Algorithmia CLI for more options.
Branches are allowed and will be accepted in pushes, however Algorithmia will not compile branches other than master.
Algorithmia will only compile the head of the master branch on each push. It will not compile intermediate commits if they are all pushed together as a batch. Finally, Algorithmia will reject any push which contains non-fast-forward commits. This is to prevent changing the history of an algorithm, and potentially breaking Algorithmia’s versioning.
Initializing a repository
As of today, you will need to manually create an algorithm from the web interface in order to initialize a repository. Once a repository is initialized you can clone it as shown above. The web interface is currently necessary in order to specify name and permissions for an algorithm.