Generate videos from Claude

Give Claude a video generator with one MCP connector

Claude cannot render video on its own. Connect Treza's MCP server and it can: ask Claude to run one of your pipelines and it starts the render, watches progress, and hands back the finished video's link.

Build a pipeline to run from Claude->See pricing

A remote MCP server, not a plugin

Treza exposes a hosted Model Context Protocol server at trezalabs.com/api/mcp. Add it to claude.ai as a custom connector, or to Claude Code with one command, and Claude gains tools for your actual account: your pipelines, your runs, your outputs.

Sign in, do not paste keys

The connector authenticates with OAuth. Claude sends you to Treza to sign in and approve access, and no API key ever touches the chat. Prefer keys for scripts? A scoped key in an Authorization header works too.

Five tools that map to real work

list_pipelines, get_pipeline, list_runs, get_run, and run_pipeline. Enough for Claude to find the right pipeline, start it, poll until it finishes, and report the output URLs, without any tool sprawl to confuse the model.

Long renders do not block the chat

Video generation takes minutes, so run_pipeline returns a run id immediately and the render continues on Treza's workers. Claude polls get_run and comes back when the status flips to success.

Scoped to exactly your account

Every tool call is checked against the signed-in account. Claude can only see and run pipelines you own, and API keys carry explicit pipelines:read and pipelines:run permissions you grant at creation.

The same pipelines you built on the canvas

Nothing new to configure per prompt. Whatever you wired on the node editor, Veo or Sora shots, narration, captions, publishing, is what Claude triggers. Fix the pipeline once and every chat-started run inherits it.

How it works

From idea to finished video

Step 01
Build or pick a pipeline
Start from a template like a captioned short, point it at your topic, and confirm a manual run works. This is the pipeline Claude will drive.
Step 02
Add the connector
In claude.ai, open Settings, Connectors, Add custom connector, and enter trezalabs.com/api/mcp. In Claude Code, run claude mcp add with the same URL.
Step 03
Approve access
Claude redirects you to Treza to sign in and approve the connection. Approve once and the connector stays linked to your account.
Step 04
Ask for a video
Say what you want: run the daily short about deep sea creatures, or check whether last night's scheduled run succeeded. Claude picks the tools, starts the run, and returns the output link.

Use cases

What people build with it

Can Claude generate videos?

Not by itself, and that is the searchable truth. With the Treza connector, Claude orchestrates real video models like Veo and Sora through your pipelines, which in practice makes the chat a video generator.

Run a content channel from chat

Your faceless channel already runs on a schedule. From Claude you can trigger an extra episode, check last night's run, or grab the output URL to share, without opening the dashboard.

Agents that produce media

Any MCP-capable agent, not just Claude, can call the same server. Give an agent a research step, then have it run your pipeline to turn the findings into a finished, captioned video.

Status checks in plain language

Did the 5pm run publish? What failed in last night's render? get_run returns per-node results, so Claude can answer with the failing node's actual error instead of a shrug.

Hands-off QA on outputs

Ask Claude to pull the last three runs and compare durations, statuses, and costs. The run history the dashboard shows is the same data the tools return.

Works in Claude Code and Cursor too

The server speaks standard Streamable HTTP MCP. Claude Code adds it with one command, and other MCP clients connect with the same URL and an API key header.

FAQ

Generate videos from Claude, answered

Can Claude AI generate videos?

Claude does not have a built-in video model. What it can do is drive one over the Model Context Protocol: connect the Treza MCP server and Claude gains tools to list your video pipelines, start runs, watch their progress, and return the finished video's URL. The generation itself happens on Treza with models like Veo 3.1, Sora 2 Pro, and Seedance 2.0.

What is the Treza MCP server?

It is a hosted Model Context Protocol endpoint at trezalabs.com/api/mcp. MCP is the open standard AI assistants use to call external tools. Adding the endpoint as a connector gives Claude, Claude Code, Cursor, or any MCP client five tools scoped to your Treza account: list_pipelines, get_pipeline, list_runs, get_run, and run_pipeline.

How do I connect Claude to Treza?

In claude.ai open Settings, then Connectors, then Add custom connector, and enter trezalabs.com/api/mcp as the URL. Claude walks you through signing in to Treza and approving access. In Claude Code, run claude mcp add with the same URL and complete the same sign-in when prompted.

Does Claude need my Treza API key?

No. The connector flow uses OAuth: you sign in on Treza's own page and approve the connection, and Claude receives a token scoped to your account. API keys exist as an alternative for scripts and clients that prefer a bearer header, with pipelines:read and pipelines:run permissions you grant explicitly.

What happens when Claude starts a run?

run_pipeline starts the pipeline on Treza's background workers and returns a run id immediately, because a video render takes minutes and an MCP tool call should not hang that long. Claude then polls get_run until the status leaves running, and the finished run includes per-node results and output media URLs.

Do runs started from Claude cost the same?

Yes. A run is a run: it draws from the same prepaid credit balance, shows up in the same run history, and follows the same rate and concurrency limits as runs started from the dashboard or the API. Only successful generations are charged.

How much does it cost to make generate videos from claude with Treza?

Treza runs on prepaid credits with no subscription. Each generation is charged at the model's own rate from your balance, and only successful runs are charged, so a failed generation costs nothing. A typical video generation settles around $1.06, and credit packs start at $5 and never expire.

Can I call this as an API instead of using the canvas?

Yes. Every pipeline can be published as a versioned HTTP endpoint. Call the typed /invoke endpoint with JSON in and JSON out, or point any OpenAI SDK at the OpenAI-compatible endpoint. Swap a model on a node later and the API your product calls does not change.

Do I need to know how to edit video?

No. Start from a template, change the topic, and run it. If you do want frame-level control, the timeline editor is there with multi-track video and audio, per-clip trims, fades, and volume, but nothing about the automated path requires opening it.

Keep exploring