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

# Build with Apostra

> Plan, run, automate, and report on advertising across sellers from one connected place.

Apostra helps you plan, run, and understand advertising across sellers from one
connected place. Use it directly, ask an assistant to help, automate repeatable
work, or build it into your own product.

You do not need to know that you are building an "advertising agent" or which
advertising systems you need. Start with the job: find places to advertise,
compare options, prepare and run a campaign, bring in customer or creative data,
or get results into a report. Apostra keeps the advertising records and carries
out supported work across connected sellers and providers.

For developers, Apostra provides one account-aware MCP and API connection for
those jobs. Your application supplies the interface, business data, and the
decisions it needs to own.

For example, an agent can turn a brief into a draft campaign, request proposals
from eligible sellers, compare the returned offers, stage a media buy, and ask
the customer to approve launch. A reporting service can read that campaign's
delivery without running an agent at all.

## What you can build

| Your product                             | How Apostra helps                                                                                                        |
| ---------------------------------------- | ------------------------------------------------------------------------------------------------------------------------ |
| A marketing assistant                    | Turn a goal into advertising options, prepare a campaign, keep a person in control of approval, and explain the results. |
| A buyer agent                            | Discover sellers, request proposals, prepare campaigns and creatives, stage buys, and launch with explicit confirmation. |
| A seller integration                     | Connect inventory, provide seller materials and business rules, handle buying work, and inspect delivery.                |
| A reporting pipeline                     | Read campaign or seller delivery, keep coverage and currency information, and export data to your own application.       |
| A creative application with media buying | Import an approved asset, find compatible inventory, and attach the exact creative to a campaign.                        |
| An AdCP test client                      | Exercise a supported sandbox workflow and inspect the resulting state without spending money.                            |

## Why connect through Apostra's MCP

The MCP endpoint exposes tools an agent can discover and call. Apostra supplies
the advertising system behind those tools: account permissions, seller
connections, shared campaign records, proposal requests, transaction execution,
and reporting. You do not need to maintain a separate buying adapter for each
supported seller underneath that integration.

Connect to one account-resolved endpoint:

```text theme={null}
https://api.interchange.io/mcp/v3
```

Call `get_status` to see the active account and its readiness. Use `tools/list`
for the current input schemas. Buyer and seller work use the same connection;
authorization still follows the selected account and each resource.

MCP is a good fit when a coding agent or model needs to discover capabilities
and choose its next action. Use the supported v2 REST API when your application
needs direct HTTP requests, such as a scheduled export or event sender. Both
operate on the same account data. An external agent can use both, with the
appropriate credential for each surface.

<Note>
  Apostra's API, sign-in, and documentation URLs currently use the
  `apostra.com` domain. Use the URLs shown here; the product name does not
  change an endpoint.
</Note>

## Choose what you want to own

Use Apostra's application and an authorized assistant for a human-guided
workflow. Build your own loop when your product needs to control the interface,
schedule, business decisions, or recovery. A hybrid application keeps durable
checkpoints in code and uses an agent for decisions such as interpreting a
brief or comparing proposals.

Apostra remains the system of record for its advertisers, campaigns, creatives,
media buys, and reporting in each approach. Hosting an arbitrary customer-defined
agent or workflow is not currently a public v3 capability.

<CardGroup cols={2}>
  <Card title="Choose your operating loop" icon="diagram-project" href="/v3/choose-your-loop">
    Decide which decisions, data, and runtime your application should own.
  </Card>

  <Card title="Connect in five minutes" icon="plug" href="/v3/quickstart">
    Connect an MCP client, sign in, and prove account access with a read.
  </Card>

  <Card title="Build an agent" icon="robot" href="/v3/build-an-agent">
    Move from a seller-discovery prototype to a durable buying workflow.
  </Card>

  <Card title="Bring data in" icon="arrow-right-to-bracket" href="/v3/bring-data-in">
    Choose the right route for events, creative assets, inventory, and materials.
  </Card>

  <Card title="Build a reporting pipeline" icon="chart-line" href="/v3/reporting-pipeline">
    Query delivery, export it, and schedule reliable downstream processing.
  </Card>

  <Card title="Connect and test a seller" icon="store" href="/v2/setup/v3/seller-workflows">
    Configure seller inventory and exercise the no-spend own-supply workflow.
  </Card>
</CardGroup>

## Availability and existing integrations

Every authenticated Buyer and Seller Account can connect to v3. A tool's
availability and permitted resources still depend on the account, permissions,
and any feature-specific rollout. Organisation accounts currently have shared
navigation tools rather than a domain catalog.

The v3 MCP contract does not replace v2 REST, and v2 is not deprecated. Existing
integrations need no migration to keep working. Some administration, event
ingestion, export, and lifecycle operations continue to use v2. Check the
[capability limits](/v2/setup/v3/limitations) before committing to a production
workflow, and use the [tool reference](/v2/setup/v3/tool-reference) alongside the
schemas returned by your connection.
