The best Campaign Manager 360 (CM360) skills for an AI agent
Campaign Manager 360 gets one built-in reference skill — CM360 Reporting & Analysis — covering both reading and the narrow set of writes CM360 actually allows. It teaches the agent the user-profile concept every CM360 call rides on, the exact plain-named v5 reporting vocabulary (no legacy dfa: prefix), entity read paths for campaign, placement, creative and Floodlight audits, and Stage-2 trafficking writes restricted to five collections — with the ad-server discipline of archiving instead of deleting.
What's inside CM360 Reporting & Analysis
- The user-profile concept — every CM360 API call rides on a specific user profile tied to one CM360 account; the skill encodes how the connected account resolves to the right profile automatically.
- Exact v5 reporting vocabulary — CM360's dfareporting v5 API dropped the legacy "dfa:" field prefix; using the wrong (older) naming is a common source of failed report requests the skill exists to prevent.
- Entity read paths for audits — campaign, placement, creative and Floodlight (conversion) configuration, plus change-log forensics for tracing who changed what.
- Stage-2 trafficking writes — restricted to exactly five collections (campaigns, placements, ads, creatives, advertiser landing pages); everything else, including Floodlight configuration itself, is read-only from the agent by design.
- Archive, never delete — CM360 has no delete for trafficking entities; the correct operation is archiving, and new ads are created inactive by default rather than live.
Why CM360's write surface is deliberately narrow
CM360 sits above individual buying platforms as the neutral ad-server record — which is exactly why its write surface stays narrow. Floodlight configuration is measurement infrastructure, not trafficking; letting an agent modify it would risk breaking conversion tracking across every campaign that depends on it, for the sake of a task that almost never actually needs to. The five writable collections are the trafficking layer only — see the full CM360 reporting-automation playbook for what this looks like as an actual workflow.
How it gets used, and adding your own
The skill loads automatically once a CM360 account is connected and your goal touches reporting, entity audits, or one of the five writable trafficking collections. You can pin it explicitly in the Workspace composer, or capture your own CM360 reconciliation method as a custom skill — see Skills, memory & teams for the general save/propose/import mechanism shared across every platform.
Frequently asked questions
- What does Agent Planners' CM360 skill cover?
- The user-profile concept behind every CM360 call, the exact v5 reporting vocabulary, entity read paths for campaign/placement/creative/Floodlight audits, and Stage-2 trafficking writes restricted to five collections.
- Can an AI agent delete a CM360 campaign or creative?
- No — CM360 has no delete operation for trafficking entities at all. The correct action is archiving, which the skill encodes as the default instead of attempting an unsupported delete.
- Why can't the agent edit Floodlight configuration in CM360?
- Floodlight is measurement infrastructure that every campaign's conversion tracking depends on — it's deliberately read-only from the agent, while the narrower trafficking surface (campaigns, placements, ads, creatives, landing pages) is writable.
- Does the CM360 skill handle reporting as well as writes?
- Yes — it's one skill covering both: the exact v5 reporting vocabulary and entity read paths for audits, plus the narrow trafficking write surface, rather than splitting reporting and writes into separate skills the way DV360 does.
- Can I add a custom CM360 skill for my own audit process?
- Yes — the same mechanism as every platform: ask the agent to save a proven run, approve a proposed skill card, or import a playbook from a link.