Connect AI agents to the Adobe Marketing Agent MCP server

Vendor MCP

The Adobe Marketing Agent MCP connector routes your AI agent's tool calls to Adobe Marketing Agent's own MCP server through Scalekit. Each user signs in to Adobe Marketing Agent once, and Scalekit stores and refreshes their tokens, so your agent never handles credentials. It comes with 12 tools.

Tools: 12
What they doRead · write · destructive: 3 · 8 · 13 read8 write1 destructive
Users sign in with: OAuth
OAuth app: Scalekit's or your own

What you can do

Ask for marketing insights: send plain-language questions to analyze audiences, troubleshoot journeys, and retrieve marketing insights
Set working context: switch the active organization, Experience Platform sandbox, or Customer Journey Analytics dataview
Track tasks and plans: check the status of long-running tasks and approve or reject a plan before it runs

Setup

npm install @scalekit-sdk/node dotenv

pip install scalekit-sdk-python python-dotenv

Set your credentials
Add your Scalekit credentials to your .env file. Find values in app.scalekit.com > Developers > API Credentials.
.env
```
SCALEKIT_ENVIRONMENT_URL=<your-environment-url>
SCALEKIT_CLIENT_ID=<your-client-id>
SCALEKIT_CLIENT_SECRET=<your-client-secret>
```
Create the Adobe Marketing Agent MCP connection
In AgentKit > Connections, create an Adobe Marketing Agent MCP connection. The name you give it is the connection_name your code passes. See Configure connections.
Scalekit credentials are available for Adobe Marketing Agent MCP server, so you don't need to register an OAuth app. To show your own app on the consent screen, use your own credentials instead.
Use your own OAuth app
Adobe Marketing Agent MCP uses Dynamic Client Registration (DCR) — no client ID or secret is required. The only step is creating a connection in Scalekit and authorizing your Adobe account.
1. Create a connection in Scalekit
  - In the Scalekit dashboard, go to AgentKit → Connections → Create Connection.
  - Search for Adobe Marketing Agent MCP and click Create.
  - Note the Connection name — use this as connection_name in your code (e.g., adobemarketingagentmcp).
2. Authorize your Adobe account
  
  Generate an authorization link and open it in a browser to complete the Adobe OAuth flow.
  
  The user is redirected to Adobe to sign in and grant access. Scalekit stores the token and injects it automatically into every tool call — no further configuration is needed.
  
  The user must have an active Adobe Experience Cloud account. The OAuth flow uses Adobe IMS (Identity Management System) and supports Single Sign-On if your organization has it configured.

Authorize a user and make your first call

Node.js
Python

1
import { ScalekitClient } from '@scalekit-sdk/node'
2
import 'dotenv/config'
3
import { createInterface } from 'node:readline/promises'
4

5
const scalekit = new ScalekitClient(
6
  process.env.SCALEKIT_ENVIRONMENT_URL,
7
  process.env.SCALEKIT_CLIENT_ID,
8
  process.env.SCALEKIT_CLIENT_SECRET,
9
)
10
const actions = scalekit.actions
11

12
const connector = 'adobemarketingagentmcp'
13
const identifier = 'user_123'
14

15
// Generate an authorization link for the user
16
const { link } = await actions.getAuthorizationLink({ connectionName: connector, identifier })
17
console.log('Authorize Adobe Marketing Agent MCP:', link)
18
const rl = createInterface({ input: process.stdin, output: process.stdout })
19
await rl.question('Press Enter after authorizing...')
20
rl.close()
21

22
// Make your first call
23
const result = await actions.executeTool({
24
  connector,
25
  identifier,
26
  toolName: 'adobemarketingagentmcp_core-list_tasks',
27
  toolInput: {},
28
})
29
console.log(result)

npx tsx quickstart.mts

1
import os
2
from scalekit import ScalekitClient
3
from dotenv import load_dotenv
4
load_dotenv()
5

6
scalekit_client = ScalekitClient(
7
    env_url=os.getenv("SCALEKIT_ENVIRONMENT_URL"),
8
    client_id=os.getenv("SCALEKIT_CLIENT_ID"),
9
    client_secret=os.getenv("SCALEKIT_CLIENT_SECRET"),
10
)
11
actions = scalekit_client.actions
12

13
connection_name = "adobemarketingagentmcp"
14
identifier = "user_123"
15

16
# Generate an authorization link for the user
17
link_response = actions.get_authorization_link(
18
    connection_name=connection_name,
19
    identifier=identifier,
20
)
21
print("Authorize Adobe Marketing Agent MCP:", link_response.link)
22
input("Press Enter after authorizing...")
23

24
# Make your first call
25
result = actions.execute_tool(
26
    tool_input={},
27
    tool_name="adobemarketingagentmcp_core-list_tasks",
28
    connection_name=connection_name,
29
    identifier=identifier,
30
)
31
print(result)

python quickstart.py

Each user signs in once. See Authorize a user for the full flow and statuses.

Tools

Pass the exact name to execute_tool

Try in Playground Request a tool

adobemarketingagentmcp_adobe-marketing-agent-mcp-widgetSend a natural-language query to the Adobe Marketing AI assistant to analyze audiences, troubleshoot journeys, and retrieve marketing insights.Read-only
Adobe Marketing Agent Mcp Widget
Send a natural-language query to the Adobe Marketing AI assistant to analyze audiences, troubleshoot journeys, and retrieve marketing insights.
Inputs
querystringrequired
The natural-language request or feedback message to send to the assistant.
asyncboolean
Compatibility flag; set to true to request async execution.
chat_idstring
Optional identifier to correlate this request with a chat or conversation context.
execution_modestring
Controls how the task runs; use async for long-running polling-based tasks.one of syncasync
long_runningboolean
Compatibility flag; set to true to request async execution for long-running tasks.
adobemarketingagentmcp_core-get_taskRetrieve the status and events for an async task by ID; use the cursor to poll only for new events since the last fetch.Read-only
Core Get Task
Retrieve the status and events for an async task by ID; use the cursor to poll only for new events since the last fetch.
Inputs
task_idstringrequired
The unique identifier of the async task to retrieve.
cursorinteger
Offset cursor from the previous response; use 0 to fetch all events from the beginning.default 0
adobemarketingagentmcp_core-list_tasksList all async tasks associated with the current conversation context.Read-only
Core List Tasks
List all async tasks associated with the current conversation context.
Inputs
This tool takes no inputs.
adobemarketingagentmcp_core-context-management-widgetDisplay and manage the current organization, sandbox, and dataview context, allowing the user to switch between them.Write
Core Context Management Widget
Display and manage the current organization, sandbox, and dataview context, allowing the user to switch between them.
Inputs
querystring
The natural-language request or feedback message to send to the assistant.
adobemarketingagentmcp_core-feedback-widgetShow an interactive feedback form with thumbs up/down and rating categories; falls back to text-based feedback if widgets are not supported.Write
Core Feedback Widget
Show an interactive feedback form with thumbs up/down and rating categories; falls back to text-based feedback if widgets are not supported.
Inputs
querystring
The natural-language request or feedback message to send to the assistant.
adobemarketingagentmcp_core-provide_feedbackSubmit user feedback about the AI assistant experience; automatically classifies sentiment and calls the feedback API.Write
Core Provide Feedback
Submit user feedback about the AI assistant experience; automatically classifies sentiment and calls the feedback API.
Inputs
commentstring
Optional free-text comment to accompany the feedback.
flagCategoriesarray
Categories of harmful content being reported; required when flagged is true.
flaggedboolean
Set to true if the feedback reports harmful content; triggers the flag API instead of sentiment feedback.
pickListarray
Selected feedback option labels chosen from the widget's predefined list.
querystring
The natural-language request or feedback message to send to the assistant.
sentimentstring
The user's overall sentiment rating for the interaction.one of likedislike
adobemarketingagentmcp_core-set_dataviewSet the active Customer Journey Analytics dataview for the current session.Write
Core Set Dataview
Set the active Customer Journey Analytics dataview for the current session.
Inputs
dataviewNamestringrequired
The name of the Customer Journey Analytics dataview to set as the active context.
adobemarketingagentmcp_core-set_sandboxSet the active Adobe Experience Platform sandbox for the current session.Write
Core Set Sandbox
Set the active Adobe Experience Platform sandbox for the current session.
Inputs
sandboxNamestringrequired
The technical name (one word) of the Adobe Experience Platform sandbox to set as the active context.
adobemarketingagentmcp_core-switch_orgSwitch to a different Adobe organization by exchanging the current IMS token.Write
Core Switch Org
Switch to a different Adobe organization by exchanging the current IMS token.
Inputs
org_namestringrequired
The display name or IMS Org ID of the Adobe organization to switch to.
adobemarketingagentmcp_core-switch_sandbox_dataviewUpdate the active sandbox and/or dataview for the session in a single call.Write
Core Switch Sandbox Dataview
Update the active sandbox and/or dataview for the session in a single call.
Inputs
dataviewNamestring
The name of the Customer Journey Analytics dataview to set as the active context.
sandboxNamestring
The technical name (one word) of the Adobe Experience Platform sandbox to set as the active context.
adobemarketingagentmcp_core-user_preferencesRead or clear the user's persisted preferences including sandbox, dataview, org, and region settings.Write
Core User Preferences
Read or clear the user's persisted preferences including sandbox, dataview, org, and region settings.
Inputs
actionstring
Action to perform on preferences; get returns current settings, clear removes all saved preferences.one of getcleardefault get
adobemarketingagentmcp_core-plan_completion_decisionSubmit the user's approval or rejection for a pending plan before it is executed.Destructive
Core Plan Completion Decision
Submit the user's approval or rejection for a pending plan before it is executed.
Inputs
decisionstringrequired
The user's approval decision for the pending plan.one of ApproveReject

Workflows

Send a query to the Adobe Marketing AI assistant

Use adobemarketingagentmcp_adobe-marketing-agent-mcp-widget to ask questions about your campaigns, audiences, journeys, and Analytics data in plain English.

Node.js
Python

const result = await actions.executeTool({
  connector: 'adobemarketingagentmcp',
  identifier: 'user_123',
  toolName: 'adobemarketingagentmcp_adobe-marketing-agent-mcp-widget',
  toolInput: {
    query: 'What are my top performing audience segments this month?',
  },
});
console.log(result);

result = actions.execute_tool(
    connection_name="adobemarketingagentmcp",
    identifier="user_123",
    tool_name="adobemarketingagentmcp_adobe-marketing-agent-mcp-widget",
    tool_input={
        "query": "What are my top performing audience segments this month?",
    },
)
print(result)

Switch sandbox and dataview

Use adobemarketingagentmcp_core-switch_sandbox_dataview to update the active Adobe Experience Platform sandbox and Customer Journey Analytics dataview in a single call.

Node.js
Python

const result = await actions.executeTool({
  connector: 'adobemarketingagentmcp',
  identifier: 'user_123',
  toolName: 'adobemarketingagentmcp_core-switch_sandbox_dataview',
  toolInput: {
    sandboxName: 'prod',
    dataviewName: 'My Analytics View',
  },
});
console.log(result);

result = actions.execute_tool(
    connection_name="adobemarketingagentmcp",
    identifier="user_123",
    tool_name="adobemarketingagentmcp_core-switch_sandbox_dataview",
    tool_input={
        "sandboxName": "prod",
        "dataviewName": "My Analytics View",
    },
)
print(result)

Poll an async task

Some Adobe Marketing operations run asynchronously. Submit a query with execution_mode: "async", then poll with adobemarketingagentmcp_core-get_task until the task completes.

Node.js
Python

1
// Step 1 — submit async query
2
const submitted = await actions.executeTool({
3
  connector: 'adobemarketingagentmcp',
4
  identifier: 'user_123',
5
  toolName: 'adobemarketingagentmcp_adobe-marketing-agent-mcp-widget',
6
  toolInput: {
7
    query: 'Generate a full audience overlap report',
8
    execution_mode: 'async',
9
  },
10
});
11
const taskId = submitted.data?.task_id;
12

13
// Step 2 — poll until complete
14
let cursor = 0;
15
while (true) {
16
  const status = await actions.executeTool({
17
    connector: 'adobemarketingagentmcp',
18
    identifier: 'user_123',
19
    toolName: 'adobemarketingagentmcp_core-get_task',
20
    toolInput: { task_id: taskId, cursor },
21
  });
22
  cursor = status.data?.cursor ?? cursor;
23
  if (status.data?.status === 'completed') {
24
    console.log(status.data.result);
25
    break;
26
  }
27
  await new Promise(r => setTimeout(r, 2000));
28
}

1
import time
2

3
# Step 1 — submit async query
4
submitted = actions.execute_tool(
5
    connection_name="adobemarketingagentmcp",
6
    identifier="user_123",
7
    tool_name="adobemarketingagentmcp_adobe-marketing-agent-mcp-widget",
8
    tool_input={
9
        "query": "Generate a full audience overlap report",
10
        "execution_mode": "async",
11
    },
12
)
13
task_id = submitted.data.get("task_id")
14

15
# Step 2 — poll until complete
16
cursor = 0
17
while True:
18
    status = actions.execute_tool(
19
        connection_name="adobemarketingagentmcp",
20
        identifier="user_123",
21
        tool_name="adobemarketingagentmcp_core-get_task",
22
        tool_input={"task_id": task_id, "cursor": cursor},
23
    )
24
    cursor = status.data.get("cursor", cursor)
25
    if status.data.get("status") == "completed":
26
        print(status.data.get("result"))
27
        break
28
    time.sleep(2)

Read and clear user preferences

User preferences (sandbox, dataview, org, region) persist for 90 days. Use adobemarketingagentmcp_core-user_preferences to read or clear them.

Node.js
Python

1
// Read current preferences
2
const prefs = await actions.executeTool({
3
  connector: 'adobemarketingagentmcp',
4
  identifier: 'user_123',
5
  toolName: 'adobemarketingagentmcp_core-user_preferences',
6
  toolInput: { action: 'get' },
7
});
8
console.log(prefs.data);
9

10
// Clear all preferences
11
await actions.executeTool({
12
  connector: 'adobemarketingagentmcp',
13
  identifier: 'user_123',
14
  toolName: 'adobemarketingagentmcp_core-user_preferences',
15
  toolInput: { action: 'clear' },
16
});

1
# Read current preferences
2
prefs = actions.execute_tool(
3
    connection_name="adobemarketingagentmcp",
4
    identifier="user_123",
5
    tool_name="adobemarketingagentmcp_core-user_preferences",
6
    tool_input={"action": "get"},
7
)
8
print(prefs.data)
9

10
# Clear all preferences
11
actions.execute_tool(
12
    connection_name="adobemarketingagentmcp",
13
    identifier="user_123",
14
    tool_name="adobemarketingagentmcp_core-user_preferences",
15
    tool_input={"action": "clear"},
16
)

What you can do

Setup

Install the SDK

Set your credentials

Create the Adobe Marketing Agent MCP connection

Create a connection in Scalekit

Authorize your Adobe account

Authorize a user and make your first call

Tools

Call it

Returns

Adobe Marketing Agent Mcp Widget

Inputs

Core Get Task

Inputs

Core List Tasks

Inputs

Core Context Management Widget

Inputs

Core Feedback Widget

Inputs

Core Provide Feedback

Inputs

Core Set Dataview

Inputs

Core Set Sandbox

Inputs

Core Switch Org

Inputs

Core Switch Sandbox Dataview

Inputs

Core User Preferences

Inputs

Core Plan Completion Decision

Inputs

Workflows

Send a query to the Adobe Marketing AI assistant

Switch sandbox and dataview

Poll an async task

Read and clear user preferences

Related: other Marketing connectors