# Amazon Ads MCP Token Cost: Compare Tool Catalogs

Canonical page: https://www.kuudo.com/guides/amazon-ads-mcp-server-connecting-cost/

Amazon Ads MCP token cost at connection: Amazon's official server loads 92,678 tokens of tool definitions; Kuudo's loads 388 before the first call.

Author: Thomas Spicer
Updated: Oct 19, 2026

## Summary

Connecting to Amazon's official Ads MCP Server costs your agent 92,678 tokens before it does anything, almost half of a 200k window. Connecting to Kuudo's Amazon Ads MCP costs 388. And Kuudo's catalog grows without adding a token at connect.

Amazon Ads MCP token cost starts with the tool definitions loaded at connection. Amazon's official Ads MCP Server hands your agent 92,678 tokens of tool definitions the moment it connects, before it reads a single campaign. Kuudo's [Amazon Ads MCP](/features/amazon-ads-mcp/) hands it 388, with 756 Amazon Ads operations behind them. This guide covers what that gap is, why it repeats, and why it widens as a catalog grows.

Amazon Ads MCP is a Model Context Protocol server that exposes Amazon Ads API operations as tools an AI client such as Claude, ChatGPT or Cursor can call, so the assistant reads and changes campaigns through Amazon's API instead of a pasted report. Amazon's official server has been in open beta since February 2026; Kuudo's is open source and built directly on the same API. Other vendors' servers are in the [comparison of Amazon Ads MCP servers](/compare/amazon-ads-mcp-servers/).

Tokens are the meter on agent workloads: API models are priced per million of them, and every window has a ceiling. Conversation history, retrieval payloads and output length vary from run to run. A server's advertised tool catalog does not. It is a fixed cost that lands in the first request of a session and, on clients that inject the full tool list, is re-sent on every turn whether or not a tool is called.

> **Why the catalog's size comes first**
> - **Everyone has the same AI.** [ChatGPT, Claude, Gemini and Cursor](/features/ai-clients/) know nothing about your business out of the box, and your competitors use the same ones.
> - **A connection supplies the edge.** It gives the assistant your account as it is right now, your rules, and your call before anything changes.
> - **All of it shares one window with the catalog.** The catalog's claim on the window lands before your account does.

## The Amazon Ads MCP catalog is charged before your agent does any work

When an MCP client asks a server for its tools, the server answers with the definition of every tool it advertises: a name, a description, and a JSON schema for the inputs. On clients that load the full list, that answer sits in the context window from the first request on, and its size is a design decision the server made before you arrived.

| Measure | Amazon official | Kuudo Amazon Ads MCP |
|---|---|---|
| Tools advertised at connect | 110 | 3 |
| Tokens at connect | 92,678 | 388 |
| Largest single tool | 12,051 | 167 |
| Average per tool | 842.5 | 129.3 |
| Tools over 1,000 tokens | 16 | 0 |
| Share of a 200k window | 46.34% | 0.19% |
| Share of a 1M window | 9.27% | 0.04% |

Amazon's largest tool, the one that creates a target, is 12,051 tokens on its own: 31 times Kuudo's entire advertised surface. Schema makes up 94% of Amazon's catalog, ten tools carry more than half of it, and sixteen exceed 1,000 tokens each. Kuudo's Amazon Ads MCP advertises three small tools, so the [MCP](/features/mcp/) connection itself weighs almost nothing.

Everything useful arrives after the connection and competes for the same room: the campaign data a tool returns, what [Amazon Agent Atlas](/features/agent-atlas/) retrieves when an Amazon rule applies, and the judgment a [Skill](/features/skills/) carries into the run. Those are the costs worth paying. The catalog is not.

> **The AI is shared. The moat is yours.**
> - **Your account, as it is right now.** The shared AI cannot read a campaign until a connection gives it the account, and the connection's own size decides how much of the window the account gets. On a 200k window, the Amazon Ads MCP leaves 199,612 tokens of room for it; the official catalog leaves 107,322.
> - **Your judgment, running every time.** A Skill runs in whatever room the catalog leaves. One catalog takes 46% of a 200k window before the rule, the data, and the reasoning get a turn.

## On clients that resend it, the MCP catalog repeats every turn and multiplies across sub-agents

Clients that inject the full tool list send the advertised definitions on every inference call, so the connect cost is not a one-time charge that frees up. It is in every request for the life of the session. Clients that load definitions on demand pay only for what they fetch. Where the catalog is resent, two consequences follow that a bigger window does not fix.

Sub-agents multiply it. Each sub-agent context carries the catalog on its own, so ten parallel sub-agents on Amazon's server carry 926,780 tokens of tool definitions between them. On Kuudo they carry 3,880.

The constraint changes with the deployment tier and does not go away. On 128k and 200k class models, including open-weight and self-hosted deployments, 92,678 tokens is 72% and 46% of the window before any work, and capacity is the limit. On [Claude models with 1M windows](https://platform.claude.com/docs/en/about-claude/models/overview) the catalog fits at 9%, and price takes over. Carrying it uncached across a 20-turn session, estimated by pricing the cl100k_base counts at Anthropic's published base input rates:

| Model | Base input per 1M | Amazon catalog, 20 turns (est.) | Kuudo catalog, 20 turns (est.) |
|---|---|---|---|
| Claude Sonnet 5.5 | $2.00 | $3.71 | $0.016 |
| Claude Opus 5.5 | $4.00 | $7.41 | $0.031 |
| Claude Fable 5.1 | $10.00 | $18.54 | $0.078 |

Claude's own tokenizer counts text differently from the cl100k_base tokenizer used here, and a different count moves both columns. Prompt caching cuts both columns by the same multiplier, because caching lowers the price of re-sending tokens, not the number of tokens sitting in the window. Cheap tokens, no room. Plenty of room, expensive tokens. A fixed catalog loses on whichever axis your deployment cares about.

Research on long contexts adds a third cost. Chroma's context-rot study found model performance degrades as input length grows, with distractors making it worse, and the Lost in the Middle study found models miss information placed in the middle of a long input. Both are in the sources.

> **The AI is shared. The moat is yours.**
> - **Your account, as it is right now.** A fleet of agents reading your account should spend its windows on your account. Ten Kuudo sub-agents carry 3,880 tokens of definitions between them; the rest is campaigns, bids, and history.
> - **Your call, before anything changes.** Approval controls hold a proposed change for review in the same window. When the catalog has already taken half of it, the proposal, its evidence, and your review are what get squeezed.

## Kuudo's Amazon Ads MCP catalog grows without adding a token at connect

Under a static catalog, advertised breadth and connect cost are the same number. Every tool added is charged to every client in every session, called or not, at about 842 tokens on Amazon's current per-tool average. Kuudo separates the two. Its Amazon Ads MCP reaches 756 Amazon Ads API operations at a connect cost of 388 tokens, and the 757th adds nothing at connect.

| Property | Static catalog | Kuudo Amazon Ads MCP |
|---|---|---|
| Cost of each catalog addition at connect | About 842 tokens per tool | 0 tokens per operation |
| Effect of catalog growth on connect cost | Linear in advertised tools | None |
| Tools over 1,000 tokens | 16 of 110 | 0 of 3 |

> **The AI is shared. The moat is yours.**
> - **Your judgment, running every time.** Your rules grow with your business: a new campaign type, a new marketplace, a new report. Every operation those rules reach adds nothing at connect, so a Skill written next quarter runs in the same room as one written today.
> - **Your account, as it is right now.** The Amazon Ads MCP already reaches the operations for Sponsored ads, Amazon DSP (the demand-side platform), [Amazon Marketing Cloud](/features/amc/), Attribution and reporting, and your agent reaches all of them from a 388-token surface.

## What happens next: connect the lighter Amazon Ads MCP

Connect the Kuudo Amazon Ads MCP. It is the lighter server by every measure in this guide: 388 tokens at connect against 92,678, and a catalog that costs nothing at connect to grow. The window holds your campaigns instead of a catalog.

Its 756 operations reach Sponsored Products, Brands, Display and TV, Amazon DSP, Amazon Marketing Cloud (AMC) workflows and audiences, Attribution, Marketing Stream and reporting. Amazon's published tool groups do not list AMC audiences, Attribution or Marketing Stream. The [comparison with Amazon's official server](/compare/kuudo-vs-amazon-ads-mcp-server/) lays it out row by row, with every Amazon fact cited to Amazon's own pages.

The same client can hold the [Selling Partner MCP](/features/amazon-selling-partner-mcp/) and the [Vendor Central MCP](/features/amazon-vendor-central-mcp/) beside it. Inventory, orders, purchase orders and bids sit in one session, which is the whole Amazon job rather than the advertising slice of it.

- **Your cloud, your Amazon app.** The servers run in your own cloud on your own Amazon app, so calls go to Amazon and not through a vendor.
- **Your policy on every write.** Reads return data; writes wait on the policy you set, and every call is on the record.
- **Your workflows as Skills.** [Skills](/features/skills/) turn a workflow into a playbook your team can read and run from Claude, ChatGPT, Cursor, Codex, n8n or any client that speaks MCP.
- **Your results out of the window.** The [Amazon Agent Data Layer](/features/amazon-agent-flow/) returns a bounded preview and a result handle for a large query, so the context holds the evidence and not the whole table.

The method and the open items are in the reference layer below.

*Next: [why the same AI in every competitor's hands makes your account context the edge](/guides/everyone-has-the-same-ai/).*


## Frequently asked questions

### What is Amazon Ads MCP, and does Amazon have an official server?

Amazon Ads MCP is a Model Context Protocol server that exposes Amazon Ads API operations as tools an AI client can call. Amazon has an official one: the Amazon Ads MCP Server entered open beta on 2026-02-02 at advertising-ai.amazon.com/mcp for Amazon Ads partners with active API credentials, authenticating through a Login with Amazon application in token or OAuth 2.1 mode, and connecting from Claude, ChatGPT, Gemini, Kiro, Amazon Bedrock and other MCP-compatible clients.

### How many tokens does the Amazon Ads MCP Server use?

Amazon's server advertises 110 tools totaling 92,678 tokens at connect: 46% of a 200k window and 9% of a 1M window before any tool is called. Sixteen tools exceed 1,000 tokens each, and the largest is 12,051.

### Does an MCP server use tokens even when no tool is called?

On clients that inject the full tool list into every request, yes: the advertised definitions are re-sent on every turn for the life of the session. Clients that load tool definitions on demand pay the cost only for the tools they fetch. Check which mode your client uses before you size a workflow.

### Is there an alternative to Amazon's Ads MCP server?

Kuudo's Amazon Ads MCP advertises 3 tools totaling 388 tokens at connect and reaches 756 Amazon Ads API operations across 55 services behind them. It runs in your own cloud on your own Amazon app and connects from Claude, ChatGPT, Cursor or any MCP client, covering Sponsored ads, Amazon DSP, Amazon Marketing Cloud, Attribution and reporting.

### Can I run Amazon's server and Kuudo's Amazon Ads MCP at the same time?

Yes. MCP clients hold several server connections at once. Keep in mind that each connected server's advertised catalog is added to the same context window, so the connect cost of every server you attach is paid together.

### Does prompt caching remove the cost of a large tool catalog?

No. Caching lowers the price of re-sending the same tokens; it does not reduce the number of tokens occupying the window. It applies the same multiplier to both servers, so the ratio between them does not change.

### How was the 92,678 versus 388 comparison measured?

Each figure is the token count of the tool list the server returns when a client connects, counted with the cl100k_base tokenizer on 2026-09-12. Tool counts are not a coverage comparison: Amazon's 110 tools may each bundle several operations.


## Sources

- [Amazon Ads MCP Server overview](https://advertising.amazon.com/API/docs/en-us/mcp/mcp-overview)

- [Introducing the Amazon Ads MCP Server (open beta)](https://advertising.amazon.com/library/news/amazon-ads-mcp-server-open-beta)

- [Connecting to the Ads MCP Server using OAuth 2.1](https://advertising.amazon.com/API/docs/en-us/mcp/oauth)

- [Model Context Protocol specification: tools](https://modelcontextprotocol.io/specification/2025-06-18/server/tools)

- [Anthropic pricing](https://platform.claude.com/docs/en/about-claude/pricing)

- [Claude models overview](https://platform.claude.com/docs/en/about-claude/models/overview)

- [Lost in the Middle: How Language Models Use Long Contexts](https://arxiv.org/abs/2307.03172)

- [Context Rot: How Increasing Input Tokens Impacts LLM Performance](https://www.trychroma.com/research/context-rot)