How do I add an MCP server to Grok?
Open grok.com/connectors and click New Connector, then Custom. Name it Viddo, paste https://mcp.viddo.ai/mcp into the MCP Server URL field, and complete the authentication.
Grok supports custom MCP connectors. Add the Viddo server as a Custom connector, approve it in your browser, and Grok can call Viddo tools in any conversation.
Server URL
https://mcp.viddo.ai/mcpSteps
Choose Custom rather than a built-in connector or a similar entry from the catalog. Viddo is not a pre-installed service, so the Custom entry is the correct path.
Grok Business and Enterprise accounts may require a team administrator to enable or pre-provision connectors in the admin console. If the New Connector option is missing, that is the account policy.
Some configurations expose only a static credential field. Do not paste a personal access token there. Note the client version and the admin policy, and use a client build that supports native OAuth for custom connectors.
Once the Viddo MCP server is connected to Grok, your agent can generate AI video, images, and audio on your own Viddo account and credits. It reads the live model list, estimates the credit cost, and creates the job only after you approve it. This page covers Grok MCP connector, connect Grok to MCP server, remote MCP tools Grok, add MCP server Grok.
Open grok.com/connectors and click New Connector, then Custom. Name it Viddo, paste https://mcp.viddo.ai/mcp into the MCP Server URL field, and complete the authentication.
Yes. Grok discovers the tools your MCP server exposes and makes them available in conversations, the same way it handles its built-in connectors.
Yes. Grok can call the Viddo generation tools to produce video, images, and audio on your Viddo account and credits.
Some configurations expose only a static credential field. Do not paste a personal access token there. Use a client build that supports native OAuth for custom connectors, or route through a controlled backend.
Claude connects to Viddo through a custom connector. Add the remote MCP server once, approve it in your browser, and every new conversation can call Viddo tools.
Server URL
https://mcp.viddo.ai/mcpSteps
Organization admins usually add custom connectors from Organization settings, under Connectors, then Add, then Custom, then Web. Members may then be limited to connectors the organization has published, and cannot add their own.
Ask your administrator whether external MCP servers and third-party OAuth are permitted, and whether user-supplied accounts require approval.
Do not enter one. Normal users never need to fill those fields. If the interface only offers manual OAuth fields and no automatic registration, record what you see and share it with Viddo support, because that case needs a pre-registered client.
The connector is not storing its refresh token. Check the connector settings and confirm it is using OAuth sign-in rather than a fixed header. Repeated sign-in requests are not expected.
Once the Viddo MCP server is connected to Claude, your agent can generate AI video, images, and audio on your own Viddo account and credits. It reads the live model list, estimates the credit cost, and creates the job only after you approve it. This page covers Claude MCP connector, connect Claude to MCP server, remote MCP server Claude, custom connector Claude.
Open claude.ai/customize/connectors and choose Add, then Add custom connector. Name it Viddo, paste https://mcp.viddo.ai/mcp into the URL field, and continue through the authentication settings.
It is a remote server that gives Claude access to external tools and data. Viddo connects as one, so Claude can call video, image, and audio generation tools with the permissions you approve at sign-in.
No. Normal users never fill those fields. If the interface only offers manual OAuth fields and no automatic registration, that is a workspace policy case and it should be reported to Viddo support.
Team and Enterprise workspaces need an owner to add the custom connector under Organization settings, then Connectors, then Add, then Custom, then Web. Members then connect their own account to it.
Gemini can call Viddo tools directly once the Viddo MCP server is added as a custom connector. Once it is connected, you can ask for an AI video, an image, or an audio track in the same conversation where you plan the rest of your work.
Server URL
https://mcp.viddo.ai/mcpSteps
This is an account or Workspace policy setting. Check with your administrator. It is not a Viddo configuration problem.
The connector registered with a redirect address the authorization server did not accept. Close the stale tab, add the connector again, and complete the sign-in in the fresh window.
The authorization did not complete. Remove the connector, add it again, and approve the permissions in the browser window that opens. Do not substitute a business API URL or a fixed token.
Video generation takes several minutes. Ask Gemini to poll the task instead of resubmitting, so the same job is not billed twice.
Once the Viddo MCP server is connected to Gemini, your agent can generate AI video, images, and audio on your own Viddo account and credits. It reads the live model list, estimates the credit cost, and creates the job only after you approve it. This page covers Gemini MCP server, custom MCP server Gemini, Gemini connectors, setup MCP Gemini.
Open gemini.google.com/customize/connectors, choose Add, then pick the custom MCP server option. Paste https://mcp.viddo.ai/mcp into the URL field including the /mcp path, select standard MCP OAuth, save, and approve the browser sign-in.
No. Use the standard MCP OAuth flow. Do not paste a static bearer token into the URL field, because a fixed token does not refresh and the connection breaks when it expires.
AI video, images, and audio across 40+ models. Gemini reads the live model list, estimates the credit cost, and creates the job only after you confirm it.
The connector was added without completing the browser approval. Remove it, add it again, and finish the sign-in in the fresh window that opens.
Cursor connects to Viddo over remote MCP with OAuth. Add the server to your MCP configuration, click the login button, and your coding agent can call Viddo tools.
Server URL
https://mcp.viddo.ai/mcpSteps
Configuration
User or project configuration
{
"mcpServers": {
"viddo": {
"url": "https://mcp.viddo.ai/mcp"
}
}
}Check the MCP panel first to confirm the server is connected, then check the sign-in. Do not add a static bearer token to the configuration as a workaround, because it expires without any obvious error.
Your account may have approved read access only. Complete the additional write permission, then retry the same request with the same clientRequestId.
Cursor asks for confirmation before running MCP tools. This is expected behavior. Approving the Viddo tools once and enabling the trusted-server option removes the repeated prompt, subject to your own security policy.
Video generation takes several minutes. Have the agent poll the task rather than resubmitting, so the same job is not billed twice.
Once the Viddo MCP server is connected to Cursor, your agent can generate AI video, images, and audio on your own Viddo account and credits. It reads the live model list, estimates the credit cost, and creates the job only after you approve it. This page covers set up MCP in Cursor, add MCP server to Cursor, Cursor MCP configuration, Cursor mcp.json.
In Cursor go to Settings, then Customize, then MCPs, then New MCP Server to open the mcp.json configuration file. Paste the Viddo configuration below it and save. Then click the login button next to Viddo in the MCP panel and approve the sign-in in your browser.
In a .cursor/mcp.json file in your project for project-specific servers, or in ~/.cursor/mcp.json in your home directory for servers available everywhere.
The OAuth flow did not finish. Click the login button next to the Viddo entry again and complete the browser approval. A saved configuration alone is not an authorized connection.
Yes. The same Viddo tools that generate video and images are available inside your coding agent, so you can ask it to create a product clip for a landing page without leaving the editor.
Every generation draws from the credit balance of the account that authorized the connection. The web app and every MCP client share that same balance.
Estimate before you generate
1. Call viddo_get_account and show the remaining credits.
2. Call viddo_estimate_generation for the model and parameters I described.
3. Show me the estimate and wait for my confirmation before you create anything.Ask for this behavior explicitly the first time you use a new agent. It is the difference between a predictable credit bill and a surprise one.
If a job fails, the agent reports the failure and the request ID. It does not switch models or resubmit on its own, because a retry with different parameters spends credits again.
My client shows the server but no tools appear
The connection reached the server but the authorization did not complete. Disconnect the server, sign in again in the browser, and confirm the permission prompt. Do not paste a bearer token into a header field as a workaround.
The browser sign-in does not come back to my client
The client registered with a redirect address the authorization server did not accept. Close the stale tab, start the connection again from the client, and complete the sign-in in the fresh window. Reusing an old callback link always fails.
My client asks me to log in every time
The client is not storing the refresh token, or it did not request offline_access. Check the client version and its credential storage. Repeated sign-in requests are not normal behavior.
My video link no longer opens
Signed links expire. Query the task again to get a fresh URL. Do not store the old link as a permanent address.
Each client registers itself separately and holds its own tokens, so connecting several clients does not multiply your bill. Credits are shared, because every client spends from the same account balance.
To revoke one connection without touching the others, use the disconnect or logout action inside that client. If the client has no revoke action, revoke it from the Viddo account security page and then remove the server entry from the client.
Revoking takes effect immediately. The next tool call from that client fails with an authorization error, and the client has to complete a fresh browser sign-in.
Connect once and your agent gets the full Viddo catalog, from Kling and Seedance to GPT Image and Flux.
Pick the platform you use from the setup guides above and follow the steps for that client. Every platform uses the same server URL, the same OAuth sign-in, and the same Viddo credits.

Viddo AI é uma plataforma avançada de geração de imagens e vídeos com IA que permite criar vídeos e imagens impressionantes de forma rápida e fácil a partir de várias entradas. Esses modelos de IA desenvolvidos por Google, OpenAI, Grok AI, ByteDance, Alibaba, Kling, Runway, Vidu, Minimax, Elevenlabs, Midjourney e assim por diante.