Pipedream MCP Server – Connect APIs to AI Agents
Pipedream provides hosted MCP servers for its supported app integrations, letting AI agents call app-specific tools through Pipedream Connect. The server is designed for developers who want agents to act on behalf of users without building OAuth flows, token storage, or per-app API integrations themselves.
Overview
Pipedream MCP is a hosted Model Context Protocol service for connecting AI agents to app integrations and API actions managed by Pipedream. Pipedream describes the MCP server as part of Pipedream Connect, with access to thousands of supported apps and prebuilt tools through a consistent interface. It is useful when an agent needs to perform real work in connected services, such as creating tickets, updating CRM records, querying spreadsheets, sending messages, or running developer workflow actions.
What the MCP server enables
Pipedream exposes app-specific MCP servers. Each server corresponds to an app integration, such as Slack, GitHub, Google Sheets, Notion, Linear, or Gmail, and provides tools based on Pipedream's registry of prebuilt actions. For developers, the MCP server can make tool calls on behalf of end users identified by an external_user_id. If a required account has not yet been connected, Pipedream can return a Connect Link so the user can authorize the app account without the developer building the OAuth flow.
When to use it
Use Pipedream MCP when an AI product needs a broad integration layer across many apps rather than a single custom API integration. It fits SaaS copilots, support agents, workflow builders, internal operations agents, and assistants that need to act across productivity, data, CRM, messaging, developer, and customer-support tools. It is also useful when teams want managed credential storage and revocable user account connections instead of exposing app credentials directly to an AI model or client-side code.
Connection and authentication
Pipedream documents a remote MCP base URL at https://remote.mcp.pipedream.net/v3 for developers. The server supports both Streamable HTTP and SSE dynamically. Requests authenticate with a bearer access token obtained through Pipedream client-credential OAuth using a Pipedream client ID, client secret, project ID, and environment. Requests also include Pipedream-specific routing headers or query parameters such as project ID, environment, external user ID, and app slug.
Key considerations
Tool availability depends on the selected app slug, connected account, and Pipedream's registry or any custom tools published to the workspace. User credentials are stored by Pipedream and can be revoked, but agents still act with the permissions granted by connected accounts, so least-privilege app authorization and clear user consent remain important. Production use requires the appropriate Pipedream Connect setup and pricing plan, while development mode can be used to get started.
Supported Transports
streamable_http
URL: https://remote.mcp.pipedream.net/v3
sse
URL: https://remote.mcp.pipedream.net/v3
Frequently Asked Questions
- When should an AI agent use the Pipedream MCP server?
- Use it when an agent needs to work across many third-party APIs through Pipedream Connect, such as sending messages, creating tickets, updating CRM records, querying spreadsheets, or running developer workflow actions on behalf of a user.
- What does Pipedream MCP add to an AI agent's capabilities?
- It gives the agent access to app-specific tools generated from Pipedream's integration and action registry, so the agent can call live APIs rather than relying only on static model knowledge.
- What can the AI agent access or manage through Pipedream MCP?
- The agent can access tools for the selected Pipedream app slug and connected account. Available operations depend on the app integration, the connected account's permissions, and any public or custom tools available in the Pipedream workspace.
- How is authentication configured for Pipedream MCP?
- Developer requests use a bearer access token obtained with Pipedream client-credential OAuth. Requests also include the Pipedream project ID, environment, external user ID, and app slug as headers or query parameters, and user account authorization is handled through Pipedream Connect.
- Which transport should be used for Pipedream MCP?
- Pipedream documents a remote server that supports both Streamable HTTP and SSE dynamically at the same base URL. Streamable HTTP is the modern default for most MCP clients, while SSE can be used when a compatible client expects that transport.