Every agency we talk to has the same half-built tool somewhere. A client dashboard that was going to replace the Monday deck. A spreadsheet that was going to refresh itself. A Slack bot that was going to catch budget problems before 5pm. The UI took an afternoon. The data took forever, and then it broke.

That is the problem the SellerMate Public API solves. One key, and everything SellerMate can see or do for your Amazon accounts is available to whatever you build: Amazon Ads campaign and search-term performance, Amazon DSP, Amazon Marketing Cloud audiences and reports, Seller Central retail and inventory data, Vendor Central sales. Plus the actions: budgets, bids, negatives, automations and dayparting, with your workspace guardrails applied to every write.

It is included in every plan, including Free. Any workspace admin can create a key from Profile → API Keys.

Why now

A year ago an "internal tool" meant a developer, a sprint and a maintenance burden. Today it means a prompt. Agencies and brands are building their own dashboards and alerts with Claude, ChatGPT and Cursor, often by people who would never call themselves developers.

What those tools still cannot do on their own is get Amazon data. Amazon Ads API approval takes weeks. The Selling Partner API has its own application process. Reports are asynchronous, rate-limited and differently shaped for every ad type. Normalising all of that across ten client accounts is months of work that nobody wants to do twice.

SellerMate already did it, for the app and for our AI agent. The Public API is that same engine, opened up.

What you can build

A few things customers are already building, each with a ready-made prompt in our prompt library:

Client dashboards. A branded page per client with spend, sales, ACoS and the week's biggest movers, behind a link you control.
Alerts in Slack or WhatsApp. Budget pacing checks at noon, stock-out warnings before ads keep spending on an empty listing.
Spreadsheets that refresh themselves. Campaigns, targets and search terms landing in a Google Sheet every morning.
Your own AI agent. An assistant that reads live account data, proposes negatives and bid changes, and applies what your guardrails allow.
Bulk operations. Budget and bid changes across every account in one run, policy-checked, fifty items per call.

How it works

Every tool is called the same way: a POST to /public/v1/tools/{name} with a JSON body, authenticated with your key in the X-API-Key header. Call get_user_context first and it tells you which Amazon accounts the key can use. Reporting tools paginate; write tools accept up to fifty items per call and return either success or a 422 with the rule that blocked them.

1curl -s -X POST https://api.sellermate.ai/public/v1/tools/get_campaign_performance \
2  -H "X-API-Key: $SELLERMATE_API_KEY" -H "Content-Type: application/json" \
3  -d '{"start_date":"2026-10-01","end_date":"2026-10-21","page_size":100}'

There are seventy-six tools at the time of writing, grouped into Discovery, Campaigns, Targeting, Negatives, Products & Reports, Tags, Amazon DSP, Amazon Marketing Cloud, Vendor Central, Knowledge Files, Automation & Dayparting, and Insights & Alerts. The full list, with every parameter, is on the documentation page, and the API publishes its own OpenAPI spec.

Built so your AI assistant can read it

This is the part we spent the most time on. The buyer of an API in 2026 is often an AI model, not a person. Someone asks Claude "can I pull my SellerMate data into a dashboard", Claude says "check if they have an API", the person pastes a link, and from then on the model is the reader.

So the documentation page is written for the model. Every tool, parameter, type and scope is in the page's HTML, not loaded by JavaScript afterwards. One fetch of that URL teaches an assistant the whole API. The prompt on the Developer API page does exactly that: it tells your assistant to read the docs first, then describe the tool you want.

We tested it the way a customer would. We asked ChatGPT how to pull campaign performance from SellerMate. It found the page, used the right header, called get_user_context first, paginated correctly and named the metrics. That is the bar.

Guardrails are the point

An API that can change bids is only useful if you can hand it to someone without handing them every account. The Public API inherits the controls your admin already set in SellerMate:

Keys scoped to accounts. Limit a key to one client's accounts, or to read only. Revoke it and it stops on the next call.
Policy on every write. Budget and bid limits, never-pause and never-archive lists, ACoS guards. A write that clears them applies immediately. One that doesn't returns 422 and nothing reaches Amazon.
Audit-logged. Every call is recorded against the key that made it, in the same log your team already uses.
Rate-limited, not metered. 120 requests a minute per key, with a Retry-After header when you hit it. No per-call pricing to forecast.

API or MCP?

If you have connected SellerMate to Claude or ChatGPT already, you have used our MCP server. The two are complementary. MCP is for conversations: ask a question, get an answer from live data, approve an action. The API is for software you build and run yourself: dashboards, syncs, alerts, agents. Many teams use both, and they share the same guardrails and audit log.

Get started

In SellerMate, open Profile → API Keys and create a key. Store it as SELLERMATE_API_KEY.
Pick a prompt from the prompt library, or write your own, and paste it into Claude, ChatGPT or Cursor.
Run what it writes.

If you build something good, tell us. We are collecting the best internal tools customers make, and the ones that help everyone may end up in the product.

Get your API key · Read the docs · Browse the prompt library

Introducing the SellerMate Public API: your Amazon data, in whatever you build