generated: '2026-09-12'
method: searched
source: https://cloud.google.com/dialogflow/cx/docs/concept/test-case,
  https://cloud.google.com/dialogflow/es/docs/agents-prebuilt,
  https://cloud.google.com/dialogflow/quotas, https://cloud.google.com/dialogflow/pricing,
  discovery/google-dialogflow-v3.json
provider: Google Dialogflow
providerId: google-dialogflow
description: >-
  Dialogflow has no test-mode key, no separate sandbox host and no fixture values. There is one
  production API, one credential type, and every runtime call is billed. What it has instead is a
  richer thing most providers lack: a first-class, API-addressable test surface — test cases,
  continuous tests, experiments and versioned environments — that lets you rehearse an agent
  against recorded conversations without a parallel sandbox account. The distinction matters for an
  agent planning a write: there is nowhere safe to practice a detectIntent, but there is a
  documented way to practice a whole conversation.
test_mode:
  separate_host: false
  separate_credentials: false
  key_prefixes: []
  test_values: []
  note: >-
    No test/live key split exists. Credentials are ordinary Google Cloud OAuth tokens and there are
    no magic test identifiers, test cards or simulator values to record. Saying so explicitly
    matters — an absent test mode is a real property of this API, not a gap in this profile.
free_to_try:
- name: Design-time requests
  price: '0.00'
  note: Navigating the builder and listing agent resources is free of charge. Only runtime
    conversation turns are billed.
- name: Dialogflow ES Trial Edition
  note: Free edition with reduced quotas (180 text rpm, 1000 audio requests/day, 15,000/month) and
    no SLA coverage.
- name: Trial credits
  note: $600 (Flows) and $1000 (Playbooks) auto-activated for new users, expiring after 12 months.
test_surfaces:
- name: Test cases
  api: Dialogflow CX (v3)
  docs: https://cloud.google.com/dialogflow/cx/docs/concept/test-case
  operations:
  - dialogflow_projects_locations_agents_testCases_create
  - dialogflow_projects_locations_agents_testCases_run
  - dialogflow_projects_locations_agents_testCases_batchRun
  - dialogflow_projects_locations_agents_testCases_calculateCoverage
  - dialogflow_projects_locations_agents_testCases_import
  - dialogflow_projects_locations_agents_testCases_export
  - dialogflow_projects_locations_agents_testCases_results_list
  limit: 2000 test cases per agent
  note: >-
    A test case records a conversation turn sequence and its expected responses, and can be replayed
    against a version or environment. calculateCoverage reports intent, transition and route-group
    coverage — a genuinely unusual capability to expose over an API.
- name: Continuous tests
  api: Dialogflow CX (v3)
  operations:
  - dialogflow_projects_locations_agents_environments_runContinuousTest
  - dialogflow_projects_locations_agents_environments_continuousTestResults_list
  note: Runs the test-case suite against a deployed environment on a schedule, so a regression
    surfaces against real deployed config.
- name: Experiments (A/B)
  api: Dialogflow CX (v3)
  operations:
  - dialogflow_projects_locations_agents_environments_experiments_create
  - dialogflow_projects_locations_agents_environments_experiments_start
  - dialogflow_projects_locations_agents_environments_experiments_stop
  note: Splits live traffic between two flow versions inside an environment and reports on the
    variants. This is the closest thing to a staged rollout Dialogflow offers.
- name: Versions and environments
  api: Dialogflow ES (v2) and CX (v3)
  note: >-
    An environment pins a version of the agent, so "draft" and "production" are real, separately
    addressable targets on the same project. Up to 20 environments per CX agent.
- name: Agent validation
  api: Dialogflow CX (v3)
  operations:
  - dialogflow_projects_locations_agents_validate
  - dialogflow_projects_locations_agents_getValidationResult
  - dialogflow_projects_locations_agents_flows_validate
  - dialogflow_projects_locations_agents_flows_getValidationResult
  note: Static analysis of an agent's configuration. There is no per-call dry-run parameter anywhere
    in the contract; this is the nearest equivalent and it is design-time only.
- name: Simulator
  api: both
  note: Interactive console-only chat panel for trying an agent. Not API-addressable.
starter_content:
- name: Prebuilt agents
  docs: https://cloud.google.com/dialogflow/es/docs/agents-prebuilt
  note: Google-published starter agents that can be imported into a project as a working baseline.
- name: Quickstart — build an agent
  docs: https://cloud.google.com/dialogflow/es/docs/quick/build-agent
caveats:
- Every runtime detectIntent is billed, including from the console simulator — "a request is
  defined as any API call to the platform, whether direct with API usage or indirect with
  integration or console usage".
- Test-case runs are agent executions and consume quota and generative tokens like any other turn.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com