Skip to main content

​
Installation

Shell command to install the CLI
wget https://downloads.getmaxim.ai/cli/<VERSION>/<OS>/<ARCH>/maxim
Example: for linux amd64
wget https://downloads.getmaxim.ai/cli/v1/linux/amd64/maxim
Use .exe extension for windows downloads. For example
wget https://downloads.getmaxim.ai/cli/v1/windows/amd64/maxim.exe

​
Supported OS + ARCH

OSARCH
Linuxamd64
Linux386
Darwinamd64
Darwinarm64
Windowsamd64
Windows386

​
Env Variables

You will require to set these environment variables before using the CLI.
NameValue
MAXIM_API_KEYAPI Key from Settings

​
Trigger a test run

Shell command to trigger a test run
maxim test -w <workflow_id> -d <dataset_id> -e <evaluator_names separated by comma>  --json
ArgumentDescription
-wWorkflow ID or IDs. If you send more than one ID (comma separated), it will create a comparison run.
-dDataset ID
-eEvaluator names. Eg Bias,BLEU
—jsonOutput in JSON (optional)

​
Programmatic Evaluator Management

Shell command to push an evaluator
maxim evaluator push --file <file_path> --workspace <workspace_id> --name <evaluator_name>
ArgumentDescription
—filePath to evaluator code file (.py, .js, .mjs, .cjs)
—workspaceWorkspace ID
—nameEvaluator name
—response-typeResponse type (Boolean, Number, String). Default: Number
—pass-criteriaPass/fail criteria as JSON (e.g. ’{“entryLevel”:{“name”:“score”,“operator”:”>=”,“value”:0.9},“runLevel”:{“name”:“queriesPassed”,“operator”:”>=”,“value”:90}}’)
—optional-variablesComma-separated list of variable names to mark as optional (e.g. ‘context,metadata’)