How to use the Data Manager Event Ingestion agent skill

What is an agent skill and what does this one do?

An agent skill is a folder containing a SKILL.md file with instructions that an AI agent reads when a task matches the skill's description. This skill is for writing ingestion code. When it is active, the agent gets step-by-step guidance for uploading conversion and event data such as offline sales, enhanced lead conversions, click conversions and Google Analytics events through the Data Manager API.

The skill does not upload anything by itself. It is a reference and workflow guide, and the text does not contain executable code. The agent uses it to pick the right official implementation guide for each destination, link you to Google's code samples and find field mapping guides when you are migrating from an older Google API.

Check that this skill fits your task

Use this skill if you are building an integration that sends offline or online conversions to Google Ads or Floodlight, migrating existing Google Ads, Measurement Protocol or Campaign Manager 360 conversion code, or sending recommended or custom GA4 events from your server.

• Not for uploading audience member lists. The skill says that is handled by the separate data-manager-api-audience-ingestion skill.
• Not for marketers who only want to view conversion reports, since this skill is for writing ingestion code.
• Not useful without access to a Google Ads, GA4 or Floodlight account, because the code cannot send data to them.

Install the skill from the ZIP file

Download the ZIP file from this page and unzip it. The package contains SKILL.md. Then place the skill folder in one of two locations: your personal skills folder at ~/.claude/skills/, or a project's .claude/skills/ folder if you want the skill available only inside that project.

If you use the Claude apps instead of a local folder, you can upload the ZIP there under Settings, then Capabilities, then Skills.

• Personal use across projects: ~/.claude/skills/
• Single project only: the .claude/skills/ folder inside the project
• Claude apps: Settings, then Capabilities, then Skills, then upload

Set up what the skill requires

Before the agent can produce working code, you need several things. The skill does not provide them. It references a separate data-manager-api-setup skill for authentication and library installation, and that skill is not included here.

You will need Data Manager API access and credentials, a client library for your language, and the account details for your destination. The skill lists Python, Java, PHP, Node and .NET libraries.

• Data Manager API access and credentials, set up through the data-manager-api-setup skill
• A client library for your language: Python, Java, PHP, Node or .NET
• Google Ads, Google Analytics or Campaign Manager 360 account details for the destination

Trigger the skill with a natural request

You do not need special syntax. Describe the task in plain language and mention the destination and the kind of data. The skill's description covers offline conversions, enhanced conversions for leads, click conversions, Google Analytics web or app events and other event ingestion cases. Requests about audience lists should go to the audience skill instead.

If the agent does not identify the destination, the skill tells it to ask which destination the events are for before writing any code, because the destination determines valid event identifiers and requirements.

• Write a Python script that uploads offline sales conversions to Google Ads with the Data Manager API.
• Help me send enhanced conversions for leads to Google Ads from my Node backend.
• I am moving our Measurement Protocol events to the Data Manager API. Show me the field mappings.
• Add server-side GA4 events with a transaction ID using the Data Manager API.
• Send Floodlight offline conversions from our Campaign Manager 360 data.

Know what the agent will check in the code

The skill sets out a workflow for the agent to follow. It starts by identifying the use case and reading the matching implementation guide, then retrieves the code sample for your language. For migrations, it has the agent pull the full field mapping guide instead of working from memory.

The implementation checklist covers initializing the client, defining destinations with operating_account, login_account and linked_account where they apply, preparing identifiers with the utility library helpers, and building the IngestEventsRequest. It also covers a validate_only option for checking schemas without uploading, and recording the request_id.

Responses can include field_warnings for non-required fields that failed validation. Because processing is asynchronous, the skill says a successful HTTP 200 response only means the payload was received. Checking whether records succeeded requires querying request status through diagnostics. The excerpt of the skill we reviewed ends partway through that step, so read the full SKILL.md for the exact query method.

• Use validate_only to test your payload before sending real data.
• Save the request_id from every ingest call for later diagnostics.
• Treat a 200 response as receipt, not proof of success.