> ## Documentation Index
> Fetch the complete documentation index at: https://docs.interchange.io/llms.txt
> Use this file to discover all available pages before exploring further.

# One storefront. Every buyer. Every channel.

> A technical introduction to running a storefront on the Scope3 Interchange — for the publishers, SSPs, and retail media networks wiring the next twenty years of inventory.

> Coming from the [Interchange overview](/v2/introduction)? This is the storefront-side deep dive — what you expose, in the order you'll expose it.

<Note>
  "Sellers" here means sellers of media — publishers, SSPs, retail media networks.
</Note>

<CardGroup cols={2}>
  <Card title="Storefront onboarding" icon="store" href="/v2/setup/storefront-onboarding">
    Register a storefront, wire an inventory source, accept your first media buy.
  </Card>

  <Card title="MCP connector" icon="plug" href="/v2/authentication">
    OAuth into Claude or ChatGPT and operate the storefront from any agent.
  </Card>
</CardGroup>

## You've integrated this market before.

If you've sold inventory programmatically in the last decade, you've taken the same meeting fifty times. Each demand source wants its own deal-id flow, its own creative spec, its own reporting cadence, its own brand-safety vendor in the loop. You've staffed an integrations team to keep up. The team is the bottleneck.

That's not a yield problem. That's a transaction-cost problem.

## The shift.

Programmatic exposes a **bid request firehose**: rate-card every impression, evaluate every QPS spike, optimize floor prices, manage daisy chains.

Agentic exposes a **product catalog**: list what you sell, describe who it reaches, accept a brief, negotiate, deliver.

You stop being a destination for bid requests. You start being a discoverable inventory catalog that buyer agents can find, query, and transact with directly.

> Programmatic asks: what is this impression worth?
> Agentic asks: which products match this brief?
> One is an auction. The other is a sales conversation — at machine speed, at any deal size.

## What you actually integrate.

You expose a **Storefront** on the Interchange. The Storefront is your
buyer-facing home; your Merchandising Agent runs it, responds to buyer briefs,
and handles the AdCP media-buy workflow. Two surfaces, one product:

* **MCP** — three tools (`health`, `ask_about_capability`, `api_call`) drive the full Interchange from any MCP-speaking model. Manage inventory sources, products, and deals from any agent surface.
* **REST + OpenAPI** — for CLIs, servers, and your existing inventory systems. OAuth for agent connectors, API keys for servers.

A Storefront connects one or more **inventory sources**. An inventory source can be an external sales agent you already run, one of your ad servers with Interchange-managed sales-agent plumbing behind it, or another Storefront. Your Merchandising Agent speaks **AdCP**, the open protocol for agentic media buying. That's how every buyer agent on the Interchange can transact with you on day one — not just the ones you've signed paper with.

External sales-agent sources can run in two modes. In **passthrough**, buyer
briefs flow to your sales agent and it answers live. In **merchandising**, the
Interchange composes products from cached **components** your source exposes.
Successful caching means Interchange can read your raw products as stable
components — ids, pricing, formats, publisher properties or selectors, delivery
type, and execution metadata — without calling your source again while composing
a buyer proposal. If your source does not support component caching yet,
passthrough can still work; Murph reports cache health separately from live
source-call health.

AdCP moves the transaction. Interchange handles the layers the protocol can't — contracts, billing, multi-agent aggregation, governance.

## Marketplace review.

Going live and appearing in the broader buyer marketplace are separate gates.
New Storefronts start pending marketplace review. Once the Storefront opens for
transactions, Scope3 reviews it before listing it for public buyer discovery.

While review is pending, the operator can keep configuring the Storefront and
work with known partners, but buyers browsing the Interchange will not see it.
After approval, admins list the Storefront; if a test or internal Storefront
should not be public, admins can keep it hidden.

## Where this leaves you.

One storefront. Every buyer agent in the marketplace. Including the buyers who would never have built a bespoke integration with you specifically — small advertisers, niche brands, public-interest campaigns, the long tail of demand that the existing system makes too expensive to serve.

When the buyer agent's transaction cost approaches zero, the deals that economics previously made impossible become trivial. A $500 podcast sponsorship. A $2,000 local news campaign. A \$200 placement during a community event. These were always worth running. They just weren't worth the overhead. They are now.

That's the change. The rest of this site explains how to ship it.

<CardGroup cols={2}>
  <Card title="Storefront onboarding" icon="user-plus" href="/v2/setup/storefront-onboarding">
    Register your storefront, expose inventory, accept buyer demand.
  </Card>

  <Card title="Storefront API reference" icon="code" href="/v2/storefront-api-reference">
    Every REST endpoint with request/response schemas.
  </Card>

  <Card title="Philosophy primer" icon="compass" href="/v2/concepts/philosophy">
    The conceptual model behind storefronts, signals, and agent-first design.
  </Card>

  <Card title="AdCP spec" icon="github" href="https://github.com/adcontextprotocol/adcp">
    The open protocol the Interchange speaks. Apache 2.0.
  </Card>
</CardGroup>
