schema_version: "2.1"
id: marketing/chatgpt-advertising-experiment-plugin
name: ChatGPT Advertising Experiment Plugin
description: Install a task-led ChatGPT Ads experiment workflow with planning skills, read-only API access, conversion reconciliation, policy gates, and a one-command mock lab.
version: 0.2.0
released_at: "2026-09-23T00:00:00Z"
category: marketing-plugins/intelligence
tags:
  - chatgpt-ads
  - advertiser-api
  - experimentation
  - measurement
  - agentic-commerce
  - governance
license: MIT
author:
  name: AI Skills Guide Maintainers
  url: https://github.com/ai-knowledge-hub/ai-skills-guide
repository: https://github.com/ai-knowledge-hub/ai-skills-guide
runtimes:
  - codex
  - claude
  - generic
entrypoints:
  spec: plugin.json
  readme: README.md
  tests: tests/test-prompts.md
  examples_dir: examples
includes:
  skills:
    - marketing/task-map-designer
    - marketing/ai-output-eval-scorecard
    - adtech/advertising-signal-ownership-auditor
    - marketing/agent-share-of-choice-evaluator
    - agentops/ad-platform-policy-gate-designer
  tools:
    - adtech/conversion-event-reconciler
    - adtech/openai-ads-api-client
  hooks:
    - require-measurement-before-activation
provider_dependencies:
  - tool: adtech/openai-ads-api-client
    requirement: required
    access: read-only
capability_readiness:
  - id: mock-lab
    name: Mock lab
    description: Run the deterministic experiment, conversion-validation, and reconciliation workflow without credentials or network access.
    access: offline
    availability: usable-now
    limitations:
      - Uses packaged fixtures only and cannot establish provider readiness.
  - id: live-read
    name: Live read
    description: Read bounded OpenAI Ads account, campaign, ad, and insights data through the authenticated read-only client.
    access: read-only
    availability: setup-required
    requires_setup:
      - Configure and verify the included OpenAI Ads API client against the intended provider account.
    limitations:
      - Provider authentication attests the ad account rather than the individual human owner of the API key.
  - id: live-write
    name: Live write
    description: Apply a provider mutation only through exact approval, a single-use grant, governed execution, and reconciliation.
    access: read-write
    availability: not-verified
    limitations:
      - No supported OpenAI Ads mutation adapter is included, so control-plane or executor setup cannot enable provider writes.
usability:
  availability: usable-now
  execution: bundle
  requires_setup:
    - Python 3.9 or newer is required to run the offline mock lab.
  limitations:
    - The included OpenAI Ads client does not mutate campaigns, ad groups, ads, files, or feeds.
    - Live provider writes remain unavailable until a supported OpenAI Ads mutation adapter is implemented and independently verified.
  quickstart: Run `python3 scripts/run_mock_lab.py` after installation to create an offline evidence bundle.
  executable_helpers:
    - entrypoint: scripts/run_mock_lab.py
      availability: usable-now
      execution: local-tool
      limitations:
        - Requires Python 3.9 or newer and performs no network calls.
      quickstart: python3 scripts/run_mock_lab.py
execution:
  kind: bundle
  supported_platforms: [linux, macos, windows]
  supported_runtimes: [python3]
artifact:
  self_contained: true
  dependency_lock: dependencies.lock.json
  checksums: checksums.txt
  sbom: sbom.cdx.json
  provenance: provenance.json
authentication:
  status: required
  methods: [api-key]
  credential_bindings: [OPENAI_ADS_API_KEY]
  scopes: []
  setup_url: https://developers.openai.com/ads/api-reference/authentication
  credential_storage: The account-scoped API key remains in the provider tool's explicitly bound environment source and is never copied into plugin state.
  validation: The included provider tool calls GET /ad_account and binds readiness to the provider-attested account identifier.
  revocation: Revoke or rotate the account-scoped API key in OpenAI Ads Manager and remove its provider-tool binding.
verification:
  tests_min_prompts: 5
  security_reviewed: false
  evidence:
    - evidence://tests/chatgpt-advertising-experiment-plugin/mock-clean-client-lab
    - evidence://tests/chatgpt-advertising-experiment-plugin/capability-readiness-projection
    - evidence://tests/chatgpt-advertising-experiment-plugin/write-unavailable-without-provider-adapter
  last_verified_at: "2026-09-23T00:00:00Z"
deprecated: false