Skip to main content
Droplinked projects your existing catalog into the agentic surfaces AI shoppers already use — ChatGPT, Claude, Cursor, and any MCP-capable client. Listing your shop publishes it to the ACP feed, the platform-wide MCP server, and a per-merchant MCP surface at mcp.droplinked.com/{shopSlug}. No re-platforming, no new checkout. Additive distribution on top of the commerce you already run.

What MCP discovery actually is

MCP is the Model Context Protocol — the open standard agents like Claude, ChatGPT, Cursor, and Continue use to call tools at runtime. Droplinked runs an MCP server that exposes your catalog as a tool surface (search_products, find_inventory, get_product, list_shop_products, start_checkout). When an AI agent answers “find me a black wool beanie under $40,” the agent calls those tools, your inventory appears in the results, and a purchase routes through the existing storefront chokepoint.

Why opt in

Agentic traffic

Your catalog becomes a result candidate in agentic shopping sessions across every MCP-aware client.

A new commission rail

Conversions attributed to an agent settle through the 70/20/10 split — you keep 70%, the agent that surfaced you earns 10%, Droplinked takes 20% for the discovery + settlement layer.

Zero migration risk

Your storefront, PSPs, and fulfillment stack stay exactly as they are. Discovery is purely additive.

How to opt in

1

Confirm your shop is publishable

Any Droplinked shop with at least one published product flows into the ACP feed and the public MCP catalog automatically. If you connect a Shopify or other storefront, finish that onboarding first — see Connect your store or the Connect Shopify walkthrough.
2

Verify your shop appears in the ACP feed

Check the delimited artifact — it is the complete catalogue in one file, and it is the format an ACP ingester actually reads (see Feed formats):
Columns are 1-indexed: 8 is seller_name, 13 is is_eligible_checkout. If rows return, you are live in agentic discovery. If only the header comes back, see the pre-flight checklist below.
Do not run this check against https://apiv3.droplinked.com/feed/acp.json. That endpoint is live and useful, but it is cursor-paginated — a single request returns one page, so filtering it can report “not listed” for a shop that is in fact listed.
3

Verify the per-merchant MCP surface

A 200 with shop metadata + tool URLs confirms agents can pin to your shop directly. A 404 means the slug isn’t registered — confirm the slug in your dashboard.
4

Optional: advertise discovery from your storefront

Add one line to your storefront <head> so agents that arrive at your shop URL can pivot straight to your MCP surface:
Full instructions: Storefront MCP Discovery.

Catalog pre-flight checklist

Agents converge on a small set of fields when ranking candidates. Make sure each product has:
  • Title — descriptive, agent-readable (see AEO/GEO optimization)
  • Description — at least one full sentence
  • Primary image — public-accessible URL, square or 1:1 preferred
  • Price + currency — must resolve to a non-zero amount in a supported currency
  • Availability — at least one SKU with stock state in_stock or low_stock
  • Public visibility — product not gated behind a draft / unpublished flag

Pricing + commission

Listing your catalog in MCP discovery is free. You pay only on agent-attributed conversions, on a 70/20/10 split: Settlement runs on two rails:
  • USDC on Base via x402 — micropayment streaming, suitable for high-volume agent traffic. Settles continuously.
  • Fiat off-ramp via Stripe — weekly payout to your Stripe-connected bank account.
You can mix the two. See Attribution + commission for the detailed flow.

Privacy + control

The discovery surface exposes only public catalog metadata — shop name, slug, product titles, descriptions, prices, images, availability. It does not expose PII, operator data, sales volume, or PSP routing decisions. Per-merchant attestation policies (public, lenders_only, private) govern any deeper inventory + attestation surface — see Inventory MCP.
You can opt out at any time by un-publishing the shop or contacting support.

Troubleshooting

Most common causes, in order:
  1. No published products — drafts don’t appear
  2. All SKUs marked out of stock
  3. The shop slug isn’t matching seller_name — try filtering on column 9 (seller_url) containing your storefront URL instead
  4. The feed regenerates on product write plus a short refresh cycle — wait 5 minutes after publishing
  5. You checked the JSON endpoint rather than the delimited artifact and only saw the first page — re-run the check against latest.tsv.gz
Search eligibility and checkout eligibility are separate gates. Every row in the feed is search-eligible; a row only becomes checkout-eligible once the seller has published the policy URLs the ACP checkout profile requires (seller_privacy_policy, column 14, and seller_tos, column 15). Column 17, checkout_ineligible_reasons, names the exact blocker per row:
As of 2026-09-01 the feed holds 35,655 rows and 947 are checkout-eligible; the overwhelmingly dominant blocker is missing seller policy URLs, which is a dashboard fix requiring no code.
  • Verify the slug exactly matches the one in your dashboard URL
  • The slug is case-sensitive and dash-separated
  • If you recently renamed your shop, the slug may have changed — old slugs don’t redirect

Next steps

Optimize your catalog for AI

Apply answer-engine + generative-engine optimization to your titles and descriptions so agents rank you higher.

See the attribution flow

Detailed walkthrough of how an agent-driven conversion is attributed and settled.