Rate a parcel across every carrier, buy the label, track it to the door, and validate addresses before a bad one costs a return.
Shippo ships in the w6w first-party pack. It declares 14 actions, 3 health checks, and the host runs its code in a sandbox that never sees the credential.
io.w6w.shippoShippo rates a parcel across every carrier account on the connection, buys the label, tracks it to the door, and validates an address before a bad one turns into a costly return. Rating and buying are deliberately separate steps: creating a shipment only asks Shippo’s connected carriers for a price and issues nothing, while purchasing a label is what actually moves money and produces a tracking number — so a workflow that only wants a quote never risks buying one by accident.
Tracking status is read for what it actually means rather than taken at face value: an unscanned parcel reads as pre-transit or unknown rather than lost, and the two statuses that need a human’s attention — returned to sender and carrier failure — are distinct values rather than blended into everything else. Address validation runs as its own cheap step ahead of purchase, returning Shippo’s corrected version of an address so a mismatch is caught before a label is printed rather than after a parcel comes back.
Beyond shipping, this app can also refund an unused label, register or read tracking for a number bought outside Shippo, and list the carrier accounts connected to the account. Registering new carrier accounts, orders, batches, manifests and webhook management are left out, since those are dashboard-level or platform-account decisions rather than steps in an automation.
Three routes to the same 14 actions. The Workflow tab is generated from Shippo's own manifest and carries its real ids, so it is copy-pasteable; the Code and CLI examples are the same call for any action on any app, so every app-specific value in them is a blank you fill in.
address-create Store an address, optionally validating it in the same call so a bad street or apartment number is caught before a label is bought rather than after a parcel comes back.
address-validate Check whether a previously created address has every field a carrier needs, and surface any corrections Shippo found.
carrier-account-list List the carrier accounts connected to this Shippo account.
parcel-create Describe a box's dimensions and weight once, and reuse its id across shipments.
refund-create Request a refund for an unused label. Not guaranteed or instant — check transaction-get afterwards for the outcome.
shipment-create Describe a parcel and get every carrier's price for it. This BUYS NOTHING — no label, no tracking number, nothing owed. `transaction-create` is the step that spends money.
shipment-get Re-read a shipment — poll this after an async shipment-create until rates arrive.
track-create Start tracking a carrier + tracking number pair not already tracked through a Shippo transaction, and return whatever status is known so far.
transaction-create PURCHASES the label. Money moves and a tracking number is issued; refund-create is a request to the carrier, not an undo.
A workflow step names the app and the action, and the editor fills in the
connection when you pick one. This is the Step shape from the
workflow spec, carrying Shippo's real ids.
{
"manifestVersion": "2",
"name": "shippo-example",
"steps": [
{
"id": "address-create",
"uses": {
"app": "io.w6w.shippo",
"action": "address-create",
"connection": "conn_YOUR_CONNECTION_ID"
},
"with": {
"country": "<country>"
}
}
]
}address-create address-validate carrier-account-list parcel-create rate-get +9 more actions available
Every app-specific value here is a blank you have to fill in. An
app action is reached through the connection that authenticates it, so the
address is a connection id, not the app id — and connections belong to your account,
so a public page cannot know yours. Create one for Shippo, then fill in
the three blanks: conn_YOUR_CONNECTION_ID, the action key, and the
parameters that action declares. The call itself is real — the shape is transcribed
from the studio's own snippet builder, which prints the same kind of blanks — but
nothing in it is specific to Shippo. The Workflow tab is where this app's
real ids are.
npm install @w6w/sdkyarn add @w6w/sdkpnpm add @w6w/sdkdeno add npm:@w6w/sdkimport { W6wClient, isActionRun } from "@w6w/sdk";
// Reads W6W_BASE_URL and W6W_TOKEN from the environment when omitted.
const client = new W6wClient();
const envelope = await client.run({
urn: "conn_YOUR_CONNECTION_ID",
action: "address-create",
payload: {
// name: "<value>",
// company: "<value>",
// street1: "<value>",
// street2: "<value>",
// city: "<value>",
// state: "<value>",
// zip: "<value>",
country: "<value>",
// phone: "<value>",
// email: "<value>",
// isResidential: "<value>",
// validate: "<value>",
// metadata: "<value>",
},
});
if (isActionRun(envelope)) console.log(envelope.value); npm install -g @w6w/cli w6w run conn_YOUR_CONNECTION_ID --action address-create --payload '{"country":"<value>"}' Give an AI agent Shippo — without giving it Shippo's credentials. One MCP endpoint exposes every app, function and workflow the caller is entitled to, as tools it can discover and run. Access is granted per team while we onboard.
One tool call{
"name": "w6w_invoke",
"arguments": {
"ref": "app:io.w6w.shippo#address-create",
"input": {
"country": "<country>"
}
}
}
Every tool names its target with a single ref. The
app: form above doesn't name a connection at all — the
host resolves which of the caller's Shippo connections to sign
with, and refuses rather than guesses when the answer is ambiguous.
The token is attached host-side, at the moment of the call. It is never a tool argument, never in the model's context, and never in a transcript — so a prompt injection has nothing to exfiltrate.
Tools are derived per end user from what that person has actually connected and is entitled to — not one shared bot identity carrying the union of everyone's access.
Multi-step work runs on the workflow engine and returns a run handle the agent can poll — retries, branching and state survive the conversation that started them.
Shippo's declared health checks are on the surface too, so an agent can tell "the vendor is down" from "your credential expired" before it burns a retry on either.
The MCP surface is part of the hosted platform. Shippo itself is MIT, and the runtime that executes it is source-available (FSL).
Shippo declares its own checks, so its health is a property of the app rather than something the host guesses at.
Shippo's own REST API, dashboard and platform API. Carrier outages are reported by name but do not count — when FedEx is down, Shippo itself is fine and you buy a different rate.
Whether this connection's account answers, and whether its token is test or live — a test token succeeds at everything and buys nothing, which no credential check alone can see.