generated: '2026-08-13'
method: searched
source: >-
  https://docs.developers.optimizely.com/configured-commerce/docs/b2b-commerce-cloud-infrastructure-deployment,
  https://github.com/optimizely/ocp-skills/blob/main/skills/ocp-local-testing/SKILL.md,
  https://docs.developers.optimizely.com/feature-experimentation/ and
  https://docs.developers.optimizely.com/performance-edge/ testing pages
description: >-
  Optimizely does NOT publish a test/live key-prefix split, magic test values, test clocks
  or fixture triggers — there is no Stripe-style sandbox anywhere in the estate. What it
  does provide is environment separation inside the product (per-project environments with
  their own datafiles and rulesets), a provisioned Sandbox hosting environment for
  Configured Commerce, and a genuine local dev server for Optimizely Connect Platform apps.
  Nothing below is invented; where a facility does not exist it is recorded as absent.
test_live_separation:
  key_prefixes: none
  note: >-
    Optimizely personal access tokens and ODP API keys carry no test/live prefix. There is
    no way to tell from a credential whether it targets production data. This is a real gap
    for agent safety: an agent holding a token cannot infer blast radius from the token
    itself.
environments:
  model: named environments per project
  api:
    experimentation_v2: [list_environments, create_environment, get_environment, update_environment, delete_environment, get_datafile]
    flags_v1: [list_environments]
  detail: >-
    Feature Experimentation projects carry named environments (commonly `development` and
    `production`), each with its own datafile, SDK key and per-environment ruleset. A flag
    key exists independently in every environment, so enabling a ruleset is an
    environment-scoped action. This is the primary safe-testing mechanism — run against a
    non-production environment key.
  caution: >-
    `environment_key` is a path parameter, not a mode flag. There is no server-side guard
    that prevents an agent from writing to `production`.
configured_commerce:
  provisioned_environments:
  - {name: Sandbox, purpose: Customer UAT testing, provisioned_by: Optimizely}
  - {name: Production, purpose: Live commerce, provisioned_by: Optimizely}
  host_patterns:
  - {infrastructure: v1, pattern: 'project.insitesandbox.com', note: non-containerized, legacy}
  - {infrastructure: v2-3, pattern: 'project.commerce.insitesandbox.com', note: containerized, current}
  note: >-
    Development and test environments beyond the single provisioned Sandbox are the
    implementation team's responsibility; additional environments are a paid add-on. This is
    also why the published Configured Commerce OpenAPI documents carry
    servers[] = https://dogfood.commerce.insitesandbox.com — Optimizely's own sandbox tenant.
  source: https://docs.developers.optimizely.com/configured-commerce/docs/b2b-commerce-cloud-infrastructure-deployment
local_development:
  tool: ocp dev (ocp-local-env)
  product: Optimizely Connect Platform
  url: http://localhost:3000
  capability: >-
    Builds and runs an OCP app locally, watches for changes and provides a browser UI to
    exercise every component — functions, jobs, lifecycle hooks, settings forms, data sync
    sources and destinations, and Opal tools — before deploying.
  flags: ['--port <n>', '--no-open', '--path <dir>', '--config <file>']
  source: https://github.com/optimizely/ocp-skills/blob/main/skills/ocp-local-testing/SKILL.md
self_hosted:
  tool: Optimizely Agent
  repository: https://github.com/optimizely/agent
  version: v4.5.0
  note: >-
    The open-source Agent microservice can be run locally against a real SDK key, which is
    the closest thing to an offline test harness for Feature Experimentation decisioning.
    Its contract is openapi/_original/optimizely-feature-experimentation-optimizely-agent-api-openapi.json.
preview_and_qa:
- {product: Web Experimentation, facility: preview/QA mode for previewing each variation before starting an experiment}
- {product: Performance Edge, facility: 'test and troubleshoot + preview each variation', docs: 'https://docs.developers.optimizely.com/performance-edge/docs/qa-experiment'}
- {product: Campaign, facility: 'test dispatch and test message template (testmessagetemplate operation) — send a test mailing before the real send'}
- {product: 'CMS (SaaS)', facility: preview rendering for blocks}
absent:
- test card numbers or magic test identifiers
- test clocks or time simulation
- fixture or webhook-trigger tooling
- a hosted public sandbox account for the Experimentation APIs
- a mock server published alongside the OpenAPI