flagr-sdk 0.1.7

Python SDK for Flagr — OpenFeature-compatible

flagr-sdk

Python SDK for Flagr — OpenFeature-compatible feature flag client.

Connects to the Flagr Evaluation API via SSE, seeds a local in-memory cache from the initial snapshot, and applies real-time updates as flags change. Every flag evaluation is fully local — no network call per check.

Prerequisites

• Python >= 3.9
• An active Flagr account and an SDK key (sdk_live_...) from the dashboard

Install

pip install flagr-sdk

Quickstart

from flagr import FlagrProvider

provider = FlagrProvider(sdk_key="sdk_live_xxx")

# One-shot evaluation
enabled = provider.is_enabled("new-checkout", tenant_id="user-123")

# Reactive — callback fires immediately and on every flag change.
# .wait() blocks until Ctrl+C, then unsubscribes automatically.
provider.on_change(
    "new-checkout",
    tenant_id="user-123",
    callback=lambda enabled: print(f"new-checkout is now {enabled}"),
).wait()

# Or keep a reference to unsubscribe manually
sub = provider.on_change("new-checkout", tenant_id="user-123", callback=my_callback)
sub.unsubscribe()

tenant_id is any entity identifier in your product — user ID, org ID, account ID, etc.

OpenFeature

If you use the OpenFeature standard, FlagrProvider is fully compatible:

from openfeature import api
from openfeature.evaluation_context import EvaluationContext
from flagr import FlagrProvider

api.set_provider(FlagrProvider(sdk_key="sdk_live_xxx"))
client = api.get_client()

enabled = client.get_boolean_value(
    "new-checkout",
    default_value=False,
    evaluation_context=EvaluationContext(targeting_key="a1b2c3d4-e5f6-7890-abcd-ef1234567890"),
)

Running locally

# Create and activate a virtual environment
python -m venv .venv
source .venv/bin/activate

# Install the package with dev dependencies
pip install -e ".[dev]"

Running tests

# Unit tests (no server required)
pytest tests/test_provider.py -v

# Integration tests (requires a running Flagr API + valid SDK key)
FLAGR_SDK_KEY=sdk_live_xxx FLAGR_BASE_URL=http://localhost:8080 pytest tests/test_integration.py -v

# Run all tests
pytest -v

# Type-check
mypy src/

# Lint
ruff check src/ tests/

Releasing a new version

Releases are fully automated via GitLab CI using PyPI's OIDC trusted publishing — no API token required.

1. Bump the version in pyproject.toml:

   [project]
   version = "0.1.4"
   

2. Commit, tag, and push:

   git add pyproject.toml
   git commit -m "bump version to 0.1.4"
   git tag v0.1.4
   git push origin v0.1.4
   

   The tag must match the pattern v*.

3. GitLab CI runs .gitlab-ci.yml, builds with hatch, and publishes to PyPI via trusted publishing. No credentials needed locally.

PyPI setup (one-time): The PyPI project must have trusted publishing configured for the GitLab repository and the pypi GitLab CI/CD environment. See PyPI trusted publishing docs.