CONNECT APP

Build with Replicate

Run machine learning models in the cloud.

Infrastructure & Cloud

API key

Visit Replicate ↗ Start building

MCP

Give your agent Replicate tools

Every Replicate action is exposed as an MCP tool on Pipedream's remote server. Point a client at it with your end user's ID and Connect resolves that user's Replicate account for each tool call — you store no tokens.

Pipedream Connect

// accessToken: mint a short-lived token with the Connect SDK — see the MCP guide
const transport = new StreamableHTTPClientTransport(
  new URL("https://remote.mcp.pipedream.net/v3"),
  {
    requestInit: {
      headers: {
        Authorization: `Bearer ${accessToken}`,
        "x-pd-project-id": "{project_id}",
        "x-pd-environment": "production",
        "x-pd-external-user-id": "{external_user_id}", // any stable ID for this user in your system
        "x-pd-app-slug": "replicate",
      },
    },
  },
)

const mcp = new Client({ name: "my-agent", version: "1.0.0" })
await mcp.connect(transport)

const { tools } = await mcp.listTools()

// e.g. run Cancel Prediction:
const result = await mcp.callTool({
  name: "replicate-cancel-prediction",
  arguments: {
    predictionId: "Preditction Id",
  },
})

# access_token: mint a short-lived token with the Connect SDK — see the MCP guide
headers = {
    "Authorization": f"Bearer {access_token}",
    "x-pd-project-id": "{project_id}",
    "x-pd-environment": "production",
    "x-pd-external-user-id": "{external_user_id}",  # any stable ID for this user in your system
    "x-pd-app-slug": "replicate",
}

async with streamablehttp_client("https://remote.mcp.pipedream.net/v3", headers=headers) as (read, write, _):
    async with ClientSession(read, write) as session:
        await session.initialize()
        tools = await session.list_tools()

        # e.g. run Cancel Prediction:
        result = await session.call_tool("replicate-cancel-prediction", {
            "predictionId": "Preditction Id",
        })

MCP guide →

API PROXY

Call the Replicate API directly

For an endpoint with no pre-built tool, the Connect proxy forwards your request to the Replicate API with the connected user's credentials attached. You store no tokens and write no refresh logic.

Pipedream Connect

const resp = await pd.proxy.get({
  externalUserId: "{external_user_id}", // any stable ID for this user in your system
  accountId: "apn_xxxxxxx",
  url: "https://api.replicate.com/v1/predictions",
})

// Any allowed Replicate endpoint works here. Pipedream attaches the
// connected account's credentials to the outgoing request.

# The path segment is the target URL, URL-safe base64 encoded:
# https://api.replicate.com/v1/predictions

curl "https://api.pipedream.com/v1/connect/{project_id}/proxy/aHR0cHM6Ly9hcGkucmVwbGljYXRlLmNvbS92MS9wcmVkaWN0aW9ucw?external_user_id={external_user_id}&account_id=apn_xxxxxxx" \
  -H "Authorization: Bearer {access_token}" \
  -H "x-pd-environment: production"

API proxy guide →

SDK

Run Replicate actions from your backend

Connect a user's Replicate account once, then run Cancel Prediction on their behalf from your own code — TypeScript, Python, or plain HTTP.

Pipedream Connect

import { PipedreamClient } from "@pipedream/sdk"

const pd = new PipedreamClient({
  projectId: process.env.PIPEDREAM_PROJECT_ID!,
  clientId: process.env.PIPEDREAM_CLIENT_ID!,
  clientSecret: process.env.PIPEDREAM_CLIENT_SECRET!,
  projectEnvironment: "production",
})

const result = await pd.actions.run({
  id: "replicate-cancel-prediction",
  externalUserId: "{external_user_id}", // any stable ID for this user in your system
  configuredProps: {
    replicate: { authProvisionId: "apn_xxxxxxx" },
    predictionId: "Preditction Id",
  },
})

from pipedream import Pipedream

pd = Pipedream(
    client_id="{oauth_client_id}",
    client_secret="{oauth_client_secret}",
    project_id="{project_id}",
    project_environment="production",
)

result = pd.actions.run(
    id="replicate-cancel-prediction",
    external_user_id="{external_user_id}",  # any stable ID for this user in your system
    configured_props={
        "replicate": {"authProvisionId": "apn_xxxxxxx"},
        "predictionId": "Preditction Id",
    },
)

curl -X POST https://api.pipedream.com/v1/connect/{project_id}/actions/run \
  -H "Content-Type: application/json" \
  -H "X-PD-Environment: production" \
  -H "Authorization: Bearer {access_token}" \
  -d '{
    "external_user_id": "{external_user_id}",
    "id": "replicate-cancel-prediction",
    "configured_props": {
      "replicate": { "authProvisionId": "apn_xxxxxxx" },
      "predictionId": "Preditction Id"
    }
  }'