How to use the Data Manager audience ingestion agent skill

What does the skill make the agent do?

When the skill is installed, the agent helps you upload audience members, remove specific users, or clear and replace an entire audience. The skill covers Customer Match audiences and other audience use cases the Data Manager API supports, such as mobile device ID audiences.

The skill changes how the agent works in three ways. It asks which destination account you are targeting before writing any code. It points you to the official guide and code sample for that destination. It reminds you to normalize and hash user data and to check request status afterward, because processing is asynchronous.

• Add members: uses the ingest request for Customer Match audiences.
• Remove members: removes specific users from an audience.
• Remove all or replace: clears an entire audience so it can be refilled.
• Event and conversion uploads are not covered. The skill says those belong to the data-manager-api-event-ingestion skill.

How do you install the skill?

The skill is distributed as a ZIP file from the skill's page in this directory. Download the ZIP, then choose one of the two install routes below.

For a personal setup, unzip the download into ~/.claude/skills/. To limit the skill to one project, unzip it into the .claude/skills/ folder inside that project. If you use the Claude apps instead, upload the ZIP under Settings → Capabilities → Skills.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into .claude/skills/ in your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

What do you need before using it?

The skill assumes you already have code and API access. It does not build audiences without code. Set up access first, because the skill itself depends on other pieces.

Authentication and library installation are handled by the separate data-manager-api-setup skill. You also need a Google Ads or Display & Video 360 account with credentials, and one of the supported client libraries: Python, Java, PHP, Node, or .NET. Normalizing and hashing user identifiers uses a utility library.

• Data Manager API access and client libraries (via the data-manager-api-setup skill)
• Client library for Python, Java, PHP, Node, or .NET
• Google Ads or Display & Video 360 account with credentials
• Utility library for normalizing and hashing user identifiers

How do you trigger the skill with a natural request?

You do not need special syntax. Describe the task and name the destination. The skill's instructions tell the agent to stop and ask where the data is going if the destination is unclear. It will not assume Google Ads by default.

A good request says what you want to do (add, remove, or clear), which audience type, and which destination. If you have no audience yet, say so, and the agent will use the create-audience reference to find the destination ID it needs.

• Upload a CRM contact list to our Google Ads Customer Match audience.
• Remove these user emails from our DV360 Customer Match audience.
• Clear and replace all members of this Google Ads audience with a new list.
• Migrate our old Google Ads API audience upload script to the Data Manager API.

What does a correct workflow look like?

The skill sends you through a fixed sequence. First it confirms the destination, then it reads the matching implementation guide and code sample. Next it builds the destination object, formats and hashes the data, and builds the payload for the operation you chose.

Before sending, you can use validate_only to check the schema without applying changes. After sending, record the request_id and check status with retrieve_request_status. A successful response only means the payload was received, not that every record was processed.

• Confirm the destination account type (Google Ads or DV360).
• Read the upload or remove-all guide for that destination.
• Format and hash user data before building the request.
• Send with validate_only first if you want a schema check.
• Record the request_id and query retrieve_request_status.

What mistakes does the skill warn about?

The skill lists several gotchas that cause failed requests. Read them before writing code, because most are about field names and required settings.

• Hashed user identifiers need the encoding field set to HEX or BASE64.
• Uploads to a Customer Match audience require terms_of_service on the request.
• Set the address field on UserIdentifier only when postal_code, family_name, given_name and region_code are all present.
• product_destination_id must be a numeric string, not a resource name.
• ConsentStatus values are CONSENT_GRANTED and CONSENT_DENIED.
• UserIdentifier uses email_address and phone_number as field names.