MCP server for AI agents

Dreamlit’s Model Context Protocol (MCP) server lets AI agents inspect your Dreamlit workspace, create or update notification workflow drafts, and prepare workflows for publishing from the tools you already use. The server is hosted by Dreamlit. You do not need to install or run a local Dreamlit package.

Server URL

https://mcp.dreamlit.ai/mcp

Transport

Streamable HTTP for remote MCP clients.

Quickstart

Add Dreamlit MCP to your AI client with the server URL, authorize Dreamlit, then ask the agent to create or update workflows in plain English.

Finish Dreamlit setup

Create a Dreamlit account, connect your database, and configure the email or Slack channels your workflows will use.

Add the MCP server

Use this server URL in Claude Code, Cursor, Codex, or another remote MCP client:

https://mcp.dreamlit.ai/mcp

If your AI client can edit its own MCP settings, ask it:

Add Dreamlit MCP with this server URL: https://mcp.dreamlit.ai/mcp

Authorize Dreamlit

Complete the browser authorization flow when your client prompts you. Dreamlit will ask which workspace the client can access.

Start prompting

Ask your agent to check Dreamlit status, then describe the workflow you want:

Check Dreamlit MCP status, then create a welcome email workflow when a new user signs up.

When the agent creates, updates, or prepares a workflow, it can return a Dreamlit link so you can review the draft before publishing.

What is MCP?

MCP is an open standard that lets AI clients connect to external tools and data sources through one shared protocol. Instead of every AI client needing a custom Dreamlit integration, the client connects to Dreamlit’s MCP server and discovers the tools Dreamlit exposes. For Dreamlit, MCP means your agent can work with notification workflows using product-level actions:

Find the right workspace, project, or workflow
Understand Dreamlit concepts and project setup state
Create or update workflow drafts from plain-English prompts
Return preview links so you can inspect the result in Dreamlit
Prepare publishing and ask for explicit confirmation before anything goes live

Learn more in the official Model Context Protocol documentation.

What can Dreamlit MCP do?

Dreamlit MCP is focused on workflow authoring and safe workflow operations. It is not a low-level database API.

Capability	What it means
Inspect setup	Check the approved workspace, project setup, Dreamlit page links, table and column summaries, and prompting guidance.
List projects and workflows	Find the project or workflow your agent should work on.
Inspect workflows	Return normalized workflow structure and a Dreamlit preview URL before editing.
Create drafts	Ask Dreamlit to create a workflow draft from an outcome-oriented prompt.
Update drafts	Continue editing an existing workflow through the same draft history you see in the web app.
Prepare publish	Validate a workflow and return a publish confirmation payload with a preview URL.
Confirm publish	Publish only after the user explicitly approves the prepared publish request.
Unpublish	Disable a live workflow when the client has publish permission.
Hand off to the browser	Send you to Dreamlit when setup, unsupported edits, or manual review need the web app.

Prerequisites

Before connecting an AI client:

Create a Dreamlit account at app.dreamlit.ai.
Finish workspace and project setup in Dreamlit.
Connect a Supabase or Postgres database.
Configure any channels the workflows will need, such as email domains, brand kits, or Slack.
Use an MCP client that supports remote MCP over Streamable HTTP, or one that can send MCP requests through an HTTP bridge.

Dreamlit MCP works against existing Dreamlit workspaces and projects. If your AI client says setup is required, open the Dreamlit link it returns and finish setup in the web app before reconnecting.

How do I connect Dreamlit MCP with OAuth?

OAuth is the best option when your MCP client supports remote server authentication. The client connects to Dreamlit, Dreamlit opens a browser authorization screen, and you approve access to one Dreamlit workspace.

Add the Dreamlit MCP server to your client

Use the server URL:

https://mcp.dreamlit.ai/mcp

Approve Dreamlit access

Your client should open a Dreamlit authorization page. Sign in, choose the workspace, review the requested scopes, and click Allow access.

Ask the client to check status

Ask your agent:

Check Dreamlit MCP status and list my projects.

Claude Code
Cursor
Codex

claude mcp add --transport http dreamlit https://mcp.dreamlit.ai/mcp

Then open Claude Code and run:

/mcp

Choose Dreamlit and complete the browser authorization flow if Claude shows it as unauthenticated. If Dreamlit is already connected, /mcp will show the server status instead.

Add a remote MCP server in Cursor settings, or add this to ~/.cursor/mcp.json for all projects. For one project only, use .cursor/mcp.json in that project.

{
  "mcpServers": {
    "dreamlit": {
      "url": "https://mcp.dreamlit.ai/mcp"
    }
  }
}

Restart Cursor or reload the window, then open Cursor settings and check the MCP server status. Complete the OAuth flow if Cursor prompts you.

Add the server, then start the OAuth flow:

codex mcp add dreamlit --url https://mcp.dreamlit.ai/mcp
codex mcp login dreamlit --scopes workflows:read,workflows:write,workflows:publish

Complete the browser authorization flow. Restart any open Codex sessions after adding the server.

If your client only receives read access, reconnect Dreamlit and approve workflow drafting or publishing when prompted. Dreamlit defaults empty OAuth scope requests to read-only access.

How do I use a personal access token (PAT)?

Use a personal access token when your client cannot complete OAuth or when you want a project-pinned credential. Dreamlit personal access tokens are:

Scoped to one project
Shown once when created
Stored hashed by Dreamlit
Revocable from project settings
Limited by the scopes you choose

Create a token

Open Dreamlit, go to Project Settings -> Project Details -> MCP Access, then click Create token.

Choose scopes

Select the smallest set of permissions your client needs.

Copy the token now

Dreamlit only shows the token once. Store it securely.

Pass the token as a Bearer token

Configure your MCP client to send:

Authorization: Bearer dlpt_...

Claude Code
JSON config
Codex

claude mcp add --transport http dreamlit https://mcp.dreamlit.ai/mcp \
  --header "Authorization: Bearer dlpt_your_token_here"

{
  "mcpServers": {
    "dreamlit": {
      "type": "http",
      "url": "https://mcp.dreamlit.ai/mcp",
      "headers": {
        "Authorization": "Bearer dlpt_your_token_here"
      }
    }
  }
}

Store the token in an environment variable:

export DREAMLIT_MCP_TOKEN="dlpt_your_token_here"

Add the server:

codex mcp add dreamlit --url https://mcp.dreamlit.ai/mcp \
  --bearer-token-env-var DREAMLIT_MCP_TOKEN

Permissions

Dreamlit MCP uses three workflow scopes.

Scope	Allows	Typical use
`workflows:read`	`get_status`, `list_projects`, `list_workflows`, `get_workflow_and_preview_url`	Inspect setup, find workflows, review drafts and published state.
`workflows:write`	`create_or_update_workflow`	Create new workflow drafts or edit existing drafts from prompts.
`workflows:publish`	`prepare_publish`, `confirm_publish`, `unpublish_workflow`	Validate, publish, schedule broadcasts, or unpublish workflows.

OAuth grants are scoped to a workspace. Personal access tokens are scoped to one project.

Tools

Your AI client discovers these tools from the server.

Tool	Required scope	Use it for
`get_status`	`workflows:read`	Learn how Dreamlit MCP works, inspect workspace/project setup, get Dreamlit page links, and retrieve prompting guidance.
`list_projects`	`workflows:read`	Find projects in the approved workspace. Project-scoped tokens only return their own project.
`list_workflows`	`workflows:read`	Find workflows in a project, optionally filtered by trigger type.
`get_workflow_and_preview_url`	`workflows:read`	Inspect draft and published workflow structure before editing or publishing.
`create_or_update_workflow`	`workflows:write`	Create or update a draft from a natural-language prompt.
`prepare_publish`	`workflows:publish`	Validate the workflow and return the exact payload needed for confirmation.
`confirm_publish`	`workflows:publish`	Publish only after the user approves the prepared publish request.
`unpublish_workflow`	`workflows:publish`	Disable a live workflow.

Dreamlit also exposes the dreamlit://mcp/guide resource. Clients that support MCP resources can read it for prompting guidance, trigger types, unsubscribe guidance, and the recommended Dreamlit MCP flow.

Recommended workflow

Ask your agent to work in this order:

Check status

Start with get_status. This gives the agent Dreamlit concepts, setup state, project context, and links back to relevant Dreamlit pages.

Find the target

Use list_projects and list_workflows if the project or workflow is unclear.

Inspect before editing

Use get_workflow_and_preview_url before changing an existing workflow.

Create or update the draft

Use create_or_update_workflow with an outcome-oriented prompt.

Review in Dreamlit

Open the returned workflow URL to inspect the draft, preview output, and make manual adjustments if needed.

Publish deliberately

Use prepare_publish first. Only call confirm_publish after you explicitly approve the prepared publish request.

Prompting Dreamlit through MCP

Dreamlit works best when you describe the notification outcome, not the internal workflow graph. Good prompts include:

The event or schedule that should start the workflow
The audience or recipient
The message goal
Data needed for targeting or personalization
Timing, timezone, and business rules
Whether unsubscribe should be enabled
Tone or brand notes when they matter

Examples:

Check Dreamlit status, then create a workflow for the main project:
when a new order row is inserted, send the customer an order confirmation
with the order number, total, and delivery estimate.

List my workflows, inspect the welcome email workflow, and update it so
users who have not created a project after 3 days get a follow-up email.
Keep unsubscribe off because this is account onboarding.

Create a repeating workflow that sends admins a weekly usage digest every
Monday at 9 AM America/New_York. Include new signups, active users, and
failed payments from the previous week.

Prepare publish for the trial reminder workflow. Show me the preview URL
and warnings before confirming anything.

Safety and control

Dreamlit MCP is designed so agents can help with workflows without making surprise production changes.

Draft changes stay draft-only until you publish.
Publishing is always a separate confirmation step with a Dreamlit preview link.
If an action needs setup or manual review, Dreamlit sends you back to the web app.
Query steps are read-only against your application data.
Access is limited by the permissions you approve. Personal access tokens are project-scoped and can be revoked anytime.

Troubleshooting

My client says Dreamlit authorization is required

Reconnect the MCP server and complete the browser authorization flow. If you are using a personal access token, confirm the token is still active and that your client is sending it as an Authorization: Bearer ... header.

My client says it needs additional permission

The current OAuth grant or personal access token does not include the required scope. Reconnect with the requested workflow access, or create a token with the missing scope.

Dreamlit says to choose a project

Your approved workspace has multiple projects and the agent did not specify one. Ask the agent to call list_projects, then retry with the right project.

Dreamlit says setup is required

Open the returned Dreamlit setup URL, finish workspace and project setup, then reconnect your MCP client.

A workflow edit returns a browser handoff

Some changes still need the Dreamlit web app. Open the returned URL, review the draft, and continue from there.

Publishing fails after a preview

The workflow may have changed since the publish preview. Ask the agent to run prepare_publish again, review the new preview and warnings, then confirm only if it still looks right.

Tools are visible but calls time out

Workflow authoring can take longer than read-only calls because Dreamlit is planning and applying a real workflow draft. Retry once. If it keeps failing, open the returned Dreamlit URL if one is present, or contact support.

Testing with MCP Inspector

You can use the MCP Inspector to verify the server connection.

Start the Inspector

npx @modelcontextprotocol/inspector

Choose Streamable HTTP

Select Streamable HTTP as the transport.

Enter the server URL

https://mcp.dreamlit.ai/mcp

Authenticate

Use OAuth if the Inspector supports it, or add a Bearer header:

Authorization: Bearer dlpt_your_token_here

Confirm the connection

Use List tools and call get_status to confirm the connection.

Next steps

Quickstart

Connect your database and create your first workflow.

Workflow basics

Learn how Dreamlit workflows are structured.

Get support

Reach the team if MCP setup blocks you.

Last validated: 2026-04-29

Get Started

Dreamlit for...

Guides

Configuration

Workflow

Step types

Troubleshooting

Resources

MCP server for AI agents

Server URL

Transport

Quickstart

What is MCP?

What can Dreamlit MCP do?

Prerequisites

How do I connect Dreamlit MCP with OAuth?

How do I use a personal access token (PAT)?

Permissions

Tools

Recommended workflow

Prompting Dreamlit through MCP

Safety and control

Troubleshooting

My client says Dreamlit authorization is required

My client says it needs additional permission

Dreamlit says to choose a project

Dreamlit says setup is required

A workflow edit returns a browser handoff

Publishing fails after a preview

Tools are visible but calls time out

Testing with MCP Inspector

Next steps

Quickstart

Workflow basics

Get support

Get Started

Dreamlit for...

Guides

Configuration

Workflow

Step types

Troubleshooting

Resources

Documentation Index

Server URL

Transport

​Quickstart

​What is MCP?

​What can Dreamlit MCP do?

​Prerequisites

​How do I connect Dreamlit MCP with OAuth?

​How do I use a personal access token (PAT)?

​Permissions

​Tools

​Recommended workflow

​Prompting Dreamlit through MCP

​Safety and control

​Troubleshooting

​My client says Dreamlit authorization is required

​My client says it needs additional permission

​Dreamlit says to choose a project

​Dreamlit says setup is required

​A workflow edit returns a browser handoff

​Publishing fails after a preview

​Tools are visible but calls time out

​Testing with MCP Inspector

​Next steps

Quickstart

Workflow basics

Get support

Quickstart

What is MCP?

What can Dreamlit MCP do?

Prerequisites

How do I connect Dreamlit MCP with OAuth?

How do I use a personal access token (PAT)?

Permissions

Tools

Recommended workflow

Prompting Dreamlit through MCP

Safety and control

Troubleshooting

My client says Dreamlit authorization is required

My client says it needs additional permission

Dreamlit says to choose a project

Dreamlit says setup is required

A workflow edit returns a browser handoff

Publishing fails after a preview

Tools are visible but calls time out

Testing with MCP Inspector

Next steps