Lovable Documentation

A chat connector (MCP server) connects one of your own tool accounts to the Lovable chat. Once connected, Lovable can look things up in that tool while it builds for you: your Notion docs, your Linear tickets, your Miro boards. You describe what you want, and Lovable grounds the result in your real content instead of guessing. Chat connections are personal. You sign in with your own account, only you can use the connection, and it does not become part of the app you publish. Chat connectors are built on MCP (Model Context Protocol), an open standard that connects AI systems like Lovable to external tools. You don’t need to know anything about MCP to use them. Lovable includes prebuilt chat connectors for popular tools and supports custom MCP servers on all plans. Workspaces can also add MCP registries, browsable directories of MCP servers that members connect as chat connectors without entering URLs.

Why use chat connectors

Chat connectors unlock two key capabilities:

  • Bring your existing context into Lovable.
    Lovable can read your team’s documentation, tickets, and design files to build prototypes and flows aligned with your standards.
  • Connect your work in Lovable back to your workflows.
    Lovable can take follow-up actions in your connected tools, such as updating ticket statuses, adding prototype links as comments, or creating new items when supported.

Lovable becomes more powerful when it understands your team’s world, eliminating guesswork and speeding up iteration. Here is what that looks like in practice:

ScenarioExample promptExample outcome
A product manager connects Linear or JiraGenerate a prototype based on tickets in the current sprint.Lovable reads issue descriptions and acceptance criteria to build aligned, testable UI.
A marketer connects Notion or ConfluenceTurn our Q4 campaign brief into a landing page.Lovable uses the document as context to generate an on-brand, editable page.
A designer connects MiroPrototype the signup flow from our whiteboard.Lovable reads flow diagrams to scaffold UI navigation and turns them into an interactive prototype your team can test.
An engineer connects n8n or an internal MCP serverBuild a workflow UI that lists users and allows deleting them.Lovable builds an app that uses live data and workflows.

How chat connectors relate to Lovable’s other MCP features

Lovable uses MCP in three different ways, each connecting a different system for a different purpose. Chat connectors bring external tools into Lovable as context while you build.

FeatureDirectionWhat it does
Chat connectors (this page)Lovable uses external toolsLovable pulls context from services such as Notion or Linear while you build
Lovable MCP serverAI tools build in LovableExternal AI tools create, edit, and manage your Lovable projects
Agent integrationsAI assistants use your appYour users’ assistants call the actions you expose in your published app

These uses also combine: a published Lovable app with an agent integration becomes a chat connector in its own workspace, listed under Lovable apps on the Connectors page. See Use your app in Lovable chats.

How chat connections work

Chat connections behave differently from the other connection types in a few important ways:

  • Personal to you. You connect with your own account, and only you can use or manage the connection. Teammates who want the same tool connect their own accounts.
  • Chat only. Lovable uses the connection while you build. It is not included in your published app, and your app’s visitors cannot reach the connected tool through it. If your published app itself needs the tool, use an app + chat connector.
  • Read first, act when supported. Lovable can read structured content such as docs, issues, and diagrams. Where the tool supports it, Lovable can also take limited actions, such as creating or updating items.
  • Available on all plans, everywhere you build. Once connected, the tool is available as context across your projects. Workspace admins can manage which chat connectors are available, described under Managing chat connectors.

Prebuilt chat connectors

Lovable includes prebuilt chat connectors for these tools. Expand one to see what connecting it unlocks:

How to add a prebuilt chat connector

To connect a prebuilt chat connector:

1

2

3

Additional setup for specific chat connectors

n8n In addition to these steps, you need to do the following in n8n:

  • Go to Settings → MCP access and toggle Enable MCP access (requires instance owner or admin permissions)
  • For each workflow you want to expose, go to Settings in the workflow editor and toggle Available in MCP.

Amplitude If you plan to analyze your Lovable-built app, make sure it is already sending data to Amplitude before using the connector. See the Amplitude Quickstart guide for setup instructions. Hex Hex requires you to provide a custom server URL when connecting. For most users, the default URL is https://app.hex.tech/mcp. If you are on a single-tenant, EU multi-tenant, or HIPAA multi-tenant instance, replace app.hex.tech with your custom Hex domain (for example, https://your-company.hex.tech/mcp or https://eu.hex.tech/mcp).

Custom MCP servers

If a tool isn’t in the catalog but offers an MCP server, you can still connect it. Custom MCP servers bring internal or third-party systems into Lovable as chat connectors, which is perfect for private APIs, internal CRMs, or custom data sources. See Connect a custom MCP server for setup steps, authentication options, and admin controls.

Managing chat connectors in your workspace

Chat connections are personal, so you review or revoke your own connections anytime in Connectors, which you can also reach from inside a project, as described in Where to find connectors. Chat connectors, including custom MCP servers, are enabled by default on all plans, including Enterprise. On Business and Enterprise plans, workspace admins and owners can turn chat connectors off for the whole workspace or disable individual ones. See Chat connector controls for the settings and where to find them.

FAQ

Read the original on docs.lovable.dev ↗