pytest-nijam

pytest plugin for Nijam, captures your test runs and ships them to the Nijam dashboard so you can track what failed, why, and where (error log + failing line), across CI runs and over time.

pytest has no traces, so, unlike the Playwright reporter, runs won't include a trace viewer. Everything else (failures, error output, the failing line, durations, and your test source) is captured.

Install

pip install pytest-nijam

The plugin auto-activates once installed (via pytest's pytest11 entry point).

Configure

Add your project ID to pytest.ini (or pyproject.toml), and provide the ingest API key via an environment variable (it's a secret, keep it out of source control):

# pytest.ini
[pytest]
nijam_project_id = 00000000-0000-0000-0000-000000000000

# pyproject.toml
[tool.pytest.ini_options]
nijam_project_id = "00000000-0000-0000-0000-000000000000"

export NIJAM_API_KEY="nij_sk_…"   # from the Nijam dashboard → Secret keys
pytest

Both the project ID and the API key can come from either the ini file or an environment variable; the environment variable wins when both are set.

Options

If nijam_api_key or nijam_project_id is missing, the plugin disables itself with a single warning, your tests run exactly as before.

CI metadata

Run context (commit, branch, PR number, CI provider/run URL, commit author) is detected automatically from GitHub Actions, GitLab CI, CircleCI, Bitbucket Pipelines, or generic GIT_* env vars, falling back to git itself. No configuration needed.

Fan-out across CI jobs

If you split your suite across several CI jobs (e.g. one job per test path) that all feed one Nijam run, set nijam_auto_complete = false (or NIJAM_AUTO_COMPLETE=false) on each job so none of them finalizes early, then complete the run once from a post-matrix step. See the docs.

Guarantees

This plugin never breaks your test run. Every hook is fail-soft: a network error, a bad key, or an unreachable API produces a [nijam] warning and nothing more, your tests still pass or fail on their own merits.

License

MIT