Skip to main content

CLI Command Reference

Now that you understand how to execute evaluations in automated CI/CD pipelines and directly from your local machine via the Hybrid Bridge, here is the complete reference for the certops CLI.

Authentication​

certops login​

Authenticates the user and caches the session token and secure tunnel configuration required for Local Bridge evaluations.

certops login --username alice@acme.com

Execution & Lifecycle​

certops run​

The primary execution command. It manages the full lifecycle of a certification run: parsing the manifest, spinning up local bridges (if applicable), triggering the SaaS evaluation engine, and polling for the final verdict.

Arguments:

  • --manifest, -f: Path to your certops.yaml manifest.
  • --host: Target host mapping. Accepts explicit mappings (target_id=url), Local Bridge ports (local:8080), or generic URLs for single-target manifests. This option can be provided multiple times.
  • --dataset, -d: Overrides the dataset ID defined within the manifest for all targets (useful when testing against a quick subset before running a massive prod dataset).
  • --timeout: Maximum seconds to wait for the run to complete before aborting the CLI watcher (default: 1800).
  • --no-wait: Triggers the run on the SaaS and immediately returns the Run ID in a fire-and-forget fashion without polling the UI.

Example: Standard CI/CD Run

certops run -f ./certops.yaml --host "https://staging.acme.com"

Example: Multi-Target Hybrid Run

certops run -f ./certops.yaml --host "retriever=https://prod.acme.com" --host "generator=local:8080"

Asset Management​

certops upload​

Uploads a dataset file (CSV or JSON) directly to the CertOps Media Hub and returns the unified Dataset ID to be referenced in your certops.yaml.

certops upload data/golden_set.csv

Output will contain the Dataset ID, row/column count, and inferred schema types.

certops tag​

Tags a historically completed certification run. This is extremely useful for manually marking an approved candidate run as the new prod baseline for future pairwise regression comparisons.

certops tag <run_uuid> prod