Built directly on the official Selling Partner API
The Amazon Vendor Central API is the vendor side of Amazon's Selling Partner API (SP-API): purchase orders, shipments, invoices, direct fulfillment, Retail Analytics and Data Kiosk. This MCP server is built directly on it. Authorize it against your own Vendor Central account and those operations are tools your AI calls from ChatGPT, Claude, Cursor or any MCP client.
Choose Managed MCP, which Kuudo hosts and operates, or run it in your own Private Cloud.
What your agents can do
Every Vendor Central function — from procurement to analytics — is now a job your AI runs for you, not a click in the portal. Whatever your operation needs done as a vendor, your agents handle it directly.
Your AI acknowledges POs, handles cancellations and change orders, and keeps procurement moving on its own.
Submit shipment requests, retrieve labels, and confirm shipments through your AI — every artifact handled in one workflow.
Invoices and credit notes filed by your AI — never double-filed, safe to retry, and audit-clean against the POs you received.
Your AI reconciles every acknowledgement, shipment request, and invoice — every recovery window watched as it happens.
Your AI pulls DF orders, acknowledges them, and keeps state in sync — fulfillment moves while your team focuses on growth.
Shipping labels, container labels, packing slips, customer invoices, and shipment confirmations — every artifact your AI needs to ship on time.
Inventory updates posted by your AI with the right bill-to address every time — your DF warehouses stay accurate in real time.
DF invoices submitted and reconciled by your AI — payment status closes itself against shipped orders.
Track DF transaction status across order, shipping, inventory, and payment workflows — one AI view, every event.
Run Data Kiosk analytics queries through your AI — the waiting and polling handled for you, analytics on the dashboard the same day.
Sales, traffic, inventory, forecasting, and net pure margin — every legacy report including real-time variants, available to your AI agents.
/ context budget
Vendor Central uses the same SP-API stack. This shared benchmark was measured on Selling Partner endpoints, not vendor-specific tools.A smaller entry point leaves more context for the task. These totals measure each connector's full entry-point definitions. Clients such as Claude can load those definitions on demand, so the total is not necessarily paid at connection. How it was measured.
Largest definition: 1,127 tokens
Largest definition: 192 tokens
Kuudo searches a prepared catalog locally, then loads the operation detail your agent needs. The catalog can grow without loading every operation into the conversation.
Measured 2026-10-22 with cl100k_base. Counts include name, description and input schema, serialized as compact sorted JSON. They exclude conversation history, server instructions and protocol wrappers.
Amazon's definitions were transcribed from Claude's view, with its client-added name prefix removed. The platform may transform text. Kuudo's definitions were captured directly from its authenticated MCP endpoint.
Both connectors expose operation discovery. Claude also deferred Amazon's entry-point definitions in the measured session. This comparison measures their full definition footprint, not actual opening-context usage or billed tokens.
Measured 2026-10-22 · cl100k_base
/ schema size
Shared SP-API benchmark: these matched schemas cover seller operations, not vendor-specific purchase orders or invoices.Across 47 matched operations, Amazon's full definitions total 37,817 tokens; Kuudo's total 19,637. Both sides include the tool name, description and full input schema. These are the combined sizes of the measured set, loaded only when requested. How it was measured.
Largest definition: 1,629 tokens
Largest definition: 1,576 tokens
Kuudo returns a compact parameter summary by default. The full input schema is available when your agent needs more detail. The comparison above uses the full definitions on both sides.
Measured 2026-10-22 with cl100k_base, using the same fields and serialization as the entry-point comparison. Each operation is counted separately. The totals are not a typical task or session cost.
The matched operations share their operation-name suffixes, top-level business parameter names and required business parameters. Amazon adds entityId and toolVersion. Nested constraints and execution equivalence were not tested; shorter descriptions and different schema detail contribute to the totals.
Amazon's schema results were transcribed from Claude's view; Kuudo's full schemas were captured directly. The two analytics metric tools and monthEndClosing had no exact match and were excluded. No seller operations were executed.
Measured 2026-10-22 · cl100k_base
Regions & markets
Acknowledge POs across all 16 EU markets in one prompt. Compare sell-through in Japan and the US side by side. Switch markets without switching tools.
Every Amazon marketplace has its own quirks — different permissions, different regions, different rules. Kuudo handles the wiring so your chat just asks the right one.
Where it fits
Your brand team asks Claude or ChatGPT to acknowledge POs, confirm shipments, file invoices, or pull yesterday's sales — without anyone opening the Vendor Central portal.
One place for every vendor code — the right marketplace every time, each code's access kept to itself, and the same workflow across every brand your agency manages.
Stand up Vendor Central operations for a new client in days, not months. Same MCP, new vendor code, same AI workflows on day one.
Build sales, traffic, inventory, and forecasting dashboards on Data Kiosk and Rapid Retail Analytics — the same MCP your agents use in production.
Coverage
Connect Kuudo as an MCP server inside the chat your team already uses. Ask for campaigns, listings, or reports in plain language. Answers come back grounded in live Amazon data.
Run multi-step goals against Amazon's APIs. Lean, efficient tools keep agents moving, so they finish the job instead of running out of room.
Drop Kuudo tools into the visual workflow builder you already operate. Trigger campaign updates, listing fixes, and report exports on a schedule or from any upstream event.
Wire Amazon data into the loop your engineers already use. Build dashboards, debug listings, and prototype agents without leaving the editor.
Choose your deployment
Your AI, your knowledge, your rules. Choose Managed MCP for velocity or Private Cloud for infrastructure control. Start with either and move between them.
Kuudo operates the MCP services. Connect your accounts and use the AI client your team already works in.
Explore Managed MCP Set up Amazon Vendor Central MCPRun Kuudo in your own cloud account, with control of the infrastructure. Agent Flow adds pipelines and a lake in your storage.
Explore Private Cloud Choose or change your deployment
claude mcp add in Claude Code. Follow the setup guide for your deployment and client.Start with Managed MCP setup or your Private Cloud connection details, then follow your client's connection reference: ChatGPT, Claude and Claude Code.
Guides
Real operator workflows, grounded in Amazon Agent Atlas, that run through this product.
Audit the Amazon Vendor Central catalog for missing ASINs and attribute mismatches, then route each finding to the right correction or suppo...
Reconcile Amazon shortage claims against invoice, PO, shipment and receipt records, then prepare the evidence for a Vendor Central dispute.
Amazon Direct Fulfillment sends a vendor order straight to the customer. Follow the order, inventory, shipment and exception checks that kee...
Amazon Vendor Central EDI is optional: the Selling Partner API also handles purchase orders, ASNs and invoices. See which route fits each pr...
Use the Amazon Retail Analytics API to load ARA sales, inventory, traffic, forecast and Net PPM reports into a warehouse that matches Vendor...
Amazon vendor chargebacks follow published clocks and thresholds. Check each purchase order decision against them before you submit, not aft...
Vendor Central vs Seller Central: in Vendor Central you sell to Amazon at a cost price it approves; in Seller Central you sell to customers...
Read Amazon glance views and conversion rate beside matching sales and inventory reports to separate traffic, units per view, and availabili...
Use an Amazon P70 forecast within an explicit forecast policy. Align weekly demand to production lead time, then net stock and committed sup...
Rank Amazon vendor inventory health by unhealthy inventory cost, then aged sellable stock, with the ARA report's condition and view kept exp...
Reconcile Amazon Retail Analytics sales by view and reporting clock. Compare ordered and shipped revenue, returns, and the matching SP-API r...
Amazon A+ content rejected? Sort listing, asset, eligibility, and no-fix conditions across affected ASINs before repairing and resubmitting...
Use Amazon A+ content automation to scan documents and applied ASINs, check listing data and assets, patch failures, and hold submissions fo...
Audit Amazon vendor listing quality across a catalog: prioritize ASINs, resolve product type schemas, and propose validated writes for polic...
Diagnose an Amazon A+ content ASIN error through the barcode, brand string, and product type before writing modules or starting another revi...
Read Amazon Retail Analytics metrics by dashboard and SP-API report: Sourceable Product OOS, ROOS, glance views, conversion, sales, and fore...
Use Amazon Net PPM analysis to rank ASIN margin deterioration and separate cost, funding, and discount pressure where the report exposes tho...
Build an Amazon PO on-time accuracy chargeback dispute with matching PO or carrier evidence, the correct filing route, and the 30-day deadli...
Frequently asked questions
The bottom line
It acknowledges the PO before the penalty, confirms the shipment, files the invoice, checks it was accepted, pulls yesterday's sales — across every marketplace, around the clock. You set the limits and make the calls that matter. Brands grow without hiring more analysts. Agencies run more vendor codes without the context-switching. Consultants stand up a new client in days, not months.