generated: '2026-08-27'
method: searched
source: >-
  https://docs.lakekeeper.io/getting-started/, https://docs.lakekeeper.io/docs/latest/api-overview/,
  https://docs.lakekeeper.io/docs/latest/bootstrap/, https://github.com/lakekeeper/lakekeeper/tree/main/examples
summary: >-
  Lakekeeper has no hosted sandbox and no test-mode credentials, because there is no vendor-operated environment to
  sandbox against. Its equivalent is a self-contained local deployment: a docker compose example that stands up the
  catalog, its Postgres, an S3-compatible object store and (in the auth examples) an IdP, so a developer gets a
  complete disposable lakehouse on localhost. There are no test keys, no test cards, and no magic identifiers to
  record — everything is a real deployment with real credentials the developer generates.
hosted_sandbox: false
test_mode_keys: false
default_endpoint: http://localhost:8181
console_ui: http://localhost:8181 (built-in Lakekeeper Console, served by the same binary)
interactive_reference: >-
  http://localhost:8181/swagger-ui/#/ — a Swagger-UI accurate to the exact version and configuration running,
  served when LAKEKEEPER__SERVE_SWAGGER_UI is true. This is the authoritative API reference for a given
  deployment; the docs-site OpenAPI documents are the released-version snapshot.
quickstart:
  steps:
  - git clone https://github.com/lakekeeper/lakekeeper.git
  - cd lakekeeper/examples/minimal
  - docker compose up
  - open http://localhost:8181/swagger-ui/#/
  source: https://docs.lakekeeper.io/docs/latest/api-overview/
deployment_options:
- name: Self-contained docker compose examples
  detail: >-
    examples/ in the repository. The minimal example needs no external services. Since v0.13.0 the compose
    examples use SeaweedFS rather than MinIO as the S3-compatible store.
- name: Kubernetes via Helm
  detail: >-
    The official chart can bundle Postgres and OpenFGA for a throwaway cluster install
    (https://lakekeeper.github.io/lakekeeper-charts).
- name: Prebuilt binary
  detail: './lakekeeper migrate && ./lakekeeper serve'
- name: From source
  detail: cargo build; `just` is the task runner.
bootstrap:
  required: true
  detail: >-
    A fresh deployment must be bootstrapped once, through the Console UI or POST /management/v1/bootstrap. The
    endpoint runs exactly once per catalog by design; `lakekeeper reopen-bootstrap` re-opens it without touching
    server_id, catalog data or existing OpenFGA tuples.
  docs: https://docs.lakekeeper.io/docs/latest/bootstrap/
fixtures:
  detail: >-
    The repository carries integration tests against Spark, PyIceberg, Trino and StarRocks, which double as
    worked examples of connecting each engine.
  docs: https://docs.lakekeeper.io/docs/latest/engines/
dry_run:
  detail: >-
    The only dry-run affordance is `lakekeeper openfga reconcile --dry-run`, which reports the OpenFGA tuple diff
    without writing. No HTTP operation accepts a dry-run parameter.