blazerops

The command-line interface and Python client for BlazerOps: author pipelines as code, push new versions, run them, move data in and out of your organisation's datawell, and call published apps from your own code.

Install

pip install blazerops
blazerops --version

Requires Python 3.11 or newer, on macOS or Linux.

Sign in

blazerops login

The CLI prints a short code and a URL. Open the URL where you are signed in to the BlazerOps app, enter the code, and pick the organisation to work in. Your token is stored in the OS keyring, never in a dotfile.

blazerops whoami     # who you are, your organisation, and the API you are talking to
blazerops doctor     # version, auth and catalog checks; exits non-zero if anything is wrong

On CI or a headless machine with no keyring, run blazerops login --print-token once and export the token as BLAZEROPS_TOKEN.

The CLI talks to https://api.blazerops.ai. Set BLAZEROPS_API_URL to point it at a different environment.

Calling a published app from Python

The same package installs the importable blazerops client. On a server that holds the app's consumer key:

from pathlib import Path
from blazerops import BlazerOps

# BLAZEROPS_API_KEY=blzk_...  BLAZEROPS_BASE_URL=https://api.blazerops.ai
with BlazerOps() as bo:
    inv = bo.invoke("my-app", "stable",
                    {"document": Path("input.pdf"), "options": {"language": "en"}},
                    subject="user-123")
    print(inv.output)

Every invoke is idempotent and retried safely. Failures raise a BlazerOpsError subclass such as RefusedError, FailedError or TimeoutError.

Documentation

Guides and the full command reference: docs.blazerops.ai.