AdCP is the open, task-first protocol that gives agents an advertising vocabulary over MCP and A2A. This track takes its three modules apart: Media Buy (discover products, create and manage a buy, measure delivery, and the pricing models), Creative (formats, sync, review, and trafficking), and Signals Activation (audiences for agents). It closes on authorized properties and the agentic supply chain, the direct extension of the ads.txt and sellers.json work in Track C. Bidcliq maintains the Prebid.org reference implementation.

1. L3H1AdCP overview and the three modules

   Prereqs: G3

   Objectives

   • Explain AdCP as a task-first layer over MCP and A2A: named tasks with agreed JSON shapes, so any buyer agent and any sales agent interoperate without custom adapters.
   • Name the three protocol modules: Media Buy, Creative, and Signals Activation.
   • Describe capability discovery (an agent asks a server what it can do) and place the October 2025 first agent-to-agent media buy in context.

   Key concepts

   Task-first protocol (named tasks, agreed JSON)The three modules: Media Buy, Creative, Signals ActivationCapability discoveryPrebid.org reference implementation

   Assessment

   List the three AdCP modules and, for each, one task an agent would call. Explain why a shared task vocabulary removes the need for a bespoke adapter per buyer-seller pair.

   Read the lesson →

   LEARN

   The next three lessons take each module apart: Media Buy (H2), Creative (H3), and Signals Activation (H4).

2. L4H2The Media Buy protocol

   Prereqs: H1, D1

   Objectives

   • Walk the media-buy lifecycle: discover products (get_products), create a buy (create_media_buy), manage it (update_media_buy, get_media_buys), and measure delivery (get_media_buy_delivery).
   • Map a product to the inventory concepts you already know (placements, formats, targeting, availability).
   • List the pricing models a buy can use: CPM, vCPM, CPC, FLAT_RATE, CPV, CPCV, and CPP.

   Key concepts

   get_products and the product catalogcreate_media_buy and the buy lifecycleget_media_buy_delivery (pacing and measurement)Pricing models (CPM, vCPM, CPC, FLAT_RATE, CPV, CPCV, CPP)

   Assessment

   Trace a single buy from get_products through create_media_buy to get_media_buy_delivery, naming what the buyer agent sends and what the sales agent returns at each step, and pick a sensible pricing model for a video goal.

   Read the lesson →

   ANALYZE

   Proposed AdCP request validator: paste a create_media_buy payload and check it against the schema.

   PRACTICE

   Proposed lab: compose a valid create_media_buy request to a given brief.

3. L3-L4H3The Creative protocol

   Prereqs: H1

   Objectives

   • Explain how an agent discovers what a seller accepts (list_creative_formats) and registers assets (sync_creatives, list_creatives).
   • Describe the review and approval workflow: format validation, brand-safety checks, and a human sign-off where required.
   • Connect creative trafficking to the ad server (for example assets flowing into GAM).

   Key concepts

   list_creative_formatssync_creatives and list_creativesCreative review, approval, and brand safetyTrafficking to the ad server

   Assessment

   Given a sales agent that supports two formats, describe the steps a buyer agent takes to get an approved creative live, and what happens when an asset fails format validation.

   Read the lesson →

   DEBUG

   Proposed runbook: a creative is rejected; trace whether it is a format, policy, or trafficking failure.

4. L4H4The Signals Activation protocol

   Prereqs: H1, D4

   Objectives

   • Explain what a signal is in AdCP: an audience or contextual segment an agent can discover and activate against inventory.
   • Connect signals to the identity and curation knowledge from Track D: better signals raise relevance and price, the same lever as match rate.
   • Describe how a buyer agent discovers available signals and activates them on a buy.

   Key concepts

   Signals (audience and contextual segments)Signal discovery and activationSignals as the agentic side of curationRelevance raises bid density and price

   Assessment

   Explain to a publisher how exposing a first-party audience as an AdCP signal could raise the price an agent pays, and why consent still gates it.

   Read the lesson →

   LEARN

   Builds on Track D identity (D4) and connects to the IAB Tech Lab Agentic Audiences work in Track I (I3).

5. L3-L4H5Authorized properties and the agentic supply chain

   Prereqs: H1, G4, C2

   Objectives

   • Explain adagents.json and list_authorized_properties: how a buyer agent confirms a sales agent is authorized to sell a given property.
   • Frame the agentic supply chain as the direct extension of ads.txt and sellers.json: the same authorization and transparency questions, one layer up.
   • Reason about trust failures: an agent selling a property it is not authorized for is the agentic unauthorized reseller.

   Key concepts

   adagents.json and list_authorized_propertiesAuthorized propertiesThe agentic supply chainTrust failures as unauthorized agent paths

   Assessment

   Given an adagents.json and a sales agent claiming a property, decide whether the claim is authorized, and explain which Track C concept this mirrors.

   Read the lesson →

   ANALYZE

   The live audit answers the same question for header bidding: who is authorized to sell this inventory. Run an audit and read the authorization checks, then map them to the agentic equivalent.

   Run it live →

   PRACTICE

   Proposed lab: author an adagents.json that authorizes a sales agent for a set of properties.