Lucas connects to Claude as an MCP (Model Context Protocol) server, letting you generate videos directly from a Claude chat conversation instead of the Lucas GUI.
Connecting Lucas to Claude
Location: Claude (web) > Customize > Connectors > Add > Custom
- Add a name for the connector
- Enter the Lucas MCP endpoint URL in the Server URL field
- Click Add
- Click Connect and log in
- Use Lucas directly in a normal Claude chat
Note: This works in Claude Chat (claude.ai in your browser), not in Claude Code.
For the full walkthrough with screenshots, see Getting Started with Lucas in Claude.
What You Can Generate via MCP
MCP-generated videos currently rely on Lucas selecting the visual approach for you. If you need to supply your own media (uploaded images, video clips, brand assets beyond a saved brand profile) or want direct control over the visual framework, use the Lucas GUI instead.
| Use case | Best fit |
|---|---|
| Quick video from a prompt, AI-selected visuals | MCP (Claude chat) |
| Video using your own uploaded media | Lucas GUI |
| Full control over shot-by-shot visual direction | Lucas GUI |
Common Questions
Q: Is this for Claude Code, or Claude Chat in the browser? A: Claude Chat (claude.ai in your browser). It does not currently work in Claude Code.
Q: Are there specific use cases or people better suited to generating videos via MCP versus the GUI, or is it just personal preference? A: It's not just preference — it depends on how much control you need over the visuals. If you're fine letting Lucas choose the visual approach, MCP is a fast, conversational way to generate a video. If you need to supply your own media or direct the shot-by-shot visual framework, the GUI is the better fit. See the comparison above.
Q: Do I need to structure my prompt in a specific way for Claude to talk to Lucas? A: No — Lucas automatically infers audience, tone, and goal from your prompt if you don't specify them. That said, giving Claude your own context first (who this is for, what the actual goal is, what tone fits) typically produces a tighter first result, since Lucas can't infer things that only come out through back-and-forth conversation.
Q: Can the Lucas MCP connector be used in Microsoft Copilot? A: In general, any agent that supports connecting to third-party custom MCP servers can work with it. Whether it works for your organization depends on your IT/CISO's permission policies for custom connectors — there can be minor differences to work through depending on the platform.
Q: What's the integration lift for customers who want to connect Lucas via MCP? A: Setup itself is quick — the same 5-step process listed above (add connector, enter URL, connect, use). Larger integration questions (e.g., whitelabeled assistants, org-wide rollout) depend on the customer's existing AI platform setup and internal approval process.
Learn More
- Getting Started with Lucas in Claude — Step-by-step connector setup with screenshots