generated: '2026-08-13'
method: searched
source: https://segmentstream.com/cli
docs: https://segmentstream.com/cli
name: SegmentStream CLI
description: >-
  A first-party command-line tool that runs the SegmentStream Measurement Engine
  warehouse-natively — identity stitching, cross-channel attribution and
  reporting execute against the customer's own BigQuery, Snowflake or Databricks
  and the results are written straight back to that warehouse, so data never
  leaves the customer's infrastructure. Configuration is a version-controlled
  config.yaml plus sources/, destinations/, transformations/ and
  attribution_models/ directories.
status: private-beta
availability:
  label: Private beta — request access
  self_serve: false
  note: >-
    The page is headed "SegmentStream CLI · Private beta" and the primary call
    to action is "Request access", so the binary is not generally available.
license_claim: >-
  The page describes it as "An open-source CLI for warehouse-native marketing
  attribution", but no public repository was found — github.com/segmentstream
  lists 13 repos and none of them is this CLI (checked 2026-08-13).
install:
  - method: shell-script
    command: curl -fsSL https://segmentstream.com/cli/install.sh | sh
    source: https://segmentstream.com/cli
    verified: false
    probe:
      url: https://segmentstream.com/cli/install.sh
      http_status: 404
      checked: '2026-08-13'
    note: >-
      The install command published on the CLI page does not resolve — the
      script it pipes to sh returns HTTP 404 ("Not Found", 9 bytes). Consistent
      with the private-beta gate, but it means the documented install path is
      currently non-functional for the public.
  - method: npm
    command: npm i -g @segmentstream/cli
    source: https://segmentstream.com/cloud.md
    verified: false
    probe:
      url: https://registry.npmjs.org/@segmentstream/cli
      http_status: 404
      checked: '2026-08-13'
    note: >-
      Shown in the SegmentStream Cloud provisioning walkthrough, explicitly
      labeled "CLI preview — commands may change before GA". The package is not
      published to npm.
commands:
  - command: segmentstream init
    group: project
    description: >-
      Scaffolds config.yaml and the folders where sources, destinations,
      transformations and attribution models live.
    output_example: >-
      Created config.yaml; created sources/ and destinations/; created
      transformations/ and attribution_models/.
  - command: segmentstream auth add <warehouse>
    group: auth
    description: >-
      Authenticates a warehouse connection. config.yaml stores only the
      credential's name; the secret goes to the OS keychain or a gitignored
      .env.
    arguments:
      - name: warehouse
        values: [bigquery, snowflake, databricks]
    auth_methods:
      - warehouse: bigquery
        method: oauth-browser
        secret_storage: none-required
      - warehouse: databricks
        method: oauth-browser
        secret_storage: os-keychain
      - warehouse: snowflake
        method: username-password
        secret_storage: gitignored-dotenv
  - command: segmentstream run
    group: pipeline
    description: >-
      Runs the full pipeline — identity stitching, attribution and reporting —
      and writes results back to the warehouse.
  - command: segmentstream run --workflow <name>
    group: pipeline
    description: "Runs a single named workflow (example shown: refresh)."
    source: https://segmentstream.com/cloud.md
    status: preview
  - command: segmentstream workspaces create <name> --region <region>
    group: cloud
    description: "Provisions a SegmentStream Cloud workspace in a region (example: eu)."
    source: https://segmentstream.com/cloud.md
    status: preview
  - command: segmentstream deploy
    group: cloud
    description: Deploys segmentstream.yml (sources and models) to a Cloud workspace.
    source: https://segmentstream.com/cloud.md
    status: preview
configuration:
  file: config.yaml
  version_controlled: true
  directories: [sources/, destinations/, transformations/, attribution_models/]
  source_types: [events, costs, conversions]
  built_in_attribution_models:
    - first_click
    - last_click
    - linear
    - position_based
    - predictive
  custom_models: >-
    A custom model is a directory under attribution_models/ containing model.sql
    and is referenced by name from config.yaml.
  warehouses: [bigquery, snowflake, databricks]
cross_links:
  packages: packages/segmentstream-packages.yml
  authentication: authentication/segmentstream-authentication.yml
notes: >-
  Distinct from the hosted platform: the CLI is the self-hosted, bring-your-own
  warehouse form of the same engine, whereas mcp/segmentstream-mcp.yml is the
  hosted agent surface. Neither install path could be verified as of
  2026-08-13 — both the shell installer and the npm package are absent — so
  every command above is recorded from the provider's own published examples
  and none was executed.