Vimeo MCP Server & CLI

Vimeo MCP server and CLI for Claude Code, Codex and AI agents. 43 tools for transcripts, bulk folder filing, showcases, chapters, captions, tags, privacy and embed presets.

One install gives you both surfaces, the same 43 tools under the same names, from the same server, so they cannot drift apart.

Vimeo MCP connects your video library to Claude, Cursor, and any other AI agent. It reads transcripts, files videos into folders in bulk, curates showcases, writes chapters and captions, and manages tags, privacy and embed presets.

There are 43 tools, and the bulk folder operations are the point of them. Vimeo moves videos between folders one at a time, so refiling a back catalog of hundreds is hundreds of drags. Here it is one call that takes up to 100 videos.

Built by Navid Moazzez. Built on Slipway, which turns one definition of each tool into the MCP server and the CLI.

Two ways to use it

Command line

vimeo-cli runs every tool as a command. Agents that run commands, like Claude Code, Codex and OpenCode, use it on their own, and you can type the same commands in a terminal, a script or a cron job:

vimeo-cli                                          # every command, one line each
vimeo-cli get-me                                   # which account and plan the token reaches
vimeo-cli search-videos --query "webinar"
vimeo-cli get-video --video-id 123456789 --json
vimeo-cli which file videos into a folder          # find the command for a task
vimeo-cli <command> --help                         # what any command takes

--confirm is the shell spelling of the confirmation deleting needs. --json gives JSON, --compact puts it on one line, --select keeps only the fields you name, and --agent turns on all of it for a script. Exit codes are 0 ok, 1 unexpected, 2 usage or a refused write, 3 not found, 4 a rejected token or a missing scope, 5 API, 7 rate limited and 10 no token, so a script branches on the number.

vimeo-cli schema <command> prints the exact JSON Schema an MCP client receives for that tool.

MCP server, for AI agents

vimeo-mcp is what Claude Code, Claude Desktop, Cursor and the rest launch. You never run it by hand:

claude mcp add vimeo -e VIMEO_PAT=your-token -- npx -y @thenavidm/vimeo-mcp-cli

In Claude Desktop, the .mcpb extension installs on a double click. Section 4 has every other client.

Related MCP server: @vidalytics/mcp

Contents

1. What you can ask it 💬

• "What's in my Vimeo library, and how much of it is unfiled?"

• "Read the transcript of the affiliate marketing workshop and write show notes."

• "Generate chapter markers for that talk from its transcript and add them."

• "Move every video with 'Workshop' in the title into the Workshops folder."

• "Which videos are set to public that shouldn't be?"

• "Build a showcase called Bootcamp 2026 and add these eight videos to it."

• "Add English captions to this video from the VTT file I just made."

• "How many plays does each video in the AI for Creators folder have?"

• "Pull the download link for the original file of video 1096473192."

• "Which of my videos have no tags?"

The one that is impossible without this is filing a back catalog. add_videos_to_folder takes up to 100 ids in a single call, so reorganizing a library of hundreds of videos is one request rather than hundreds of drags.

2. Quick install ⚡

You need Node 22 or newer, and nothing else.

npx -y @thenavidm/vimeo-mcp-cli --version

That is the whole install. npx fetches it on demand, so there is nothing to update later.

For the CLI as a command you or your agent can run anywhere, install it once:

npm install -g @thenavidm/vimeo-mcp-cli
vimeo-cli

3. Setup 🔑

You need a Vimeo personal access token.

Read this before you generate one. A token's scopes are fixed the moment it is created and cannot be changed afterwards. If you miss a box, the only fix is to generate a new token. Two are off by default and are the ones people miss:

Have an agent do it

The agent cannot sign in to Vimeo for you. Only you can create the token. What it can do is walk you through it, wire up the config and verify the connection.

Paste this into Claude Code, Cursor, or any agent with terminal access:

Help me set up the Vimeo MCP server.

1. Tell me how to create a Vimeo personal access token, and which scopes to
   tick. Then stop and wait: I will paste the token back to you.
2. Once I give you the token, add the server to my MCP client config with the
   token in the env block.
3. Run the doctor command and tell me what it says about my scopes and plan.

Or do it yourself

1. Go to https://developer.vimeo.com/apps and sign in.

2. Create an app, or open one you already have.

3. Open the app's Authentication section.

4. Generate a personal access token.

5. Tick the scopes. public, private, edit, create, interact and upload cover everyday use. Add delete for the delete tools and video_files for download links. stats only does something on a paid plan.

6. Copy the token immediately. Vimeo shows it once.

To revoke it

Go back to the same Authentication section and delete the token. It stops working at once, everywhere it is configured.

4. Connect your client 🔌

The long version, every step with what to do when one fails, is in INSTALL.md.

Claude Code

claude mcp add vimeo \
  -e VIMEO_PAT=your_token_here \
  -- npx -y @thenavidm/vimeo-mcp-cli@latest

Add --scope user to make it available in every project rather than just this one.

Claude Desktop

The short way: download the .mcpb extension from the latest release and double-click it. It carries its own dependencies, so there is no config file to edit and nothing to install first. Claude Desktop asks for your Vimeo personal access token.

The long way, if you would rather edit the config yourself:

{
  "mcpServers": {
    "vimeo": {
      "command": "npx",
      "args": ["-y", "@thenavidm/vimeo-mcp-cli@latest"],
      "env": { "VIMEO_PAT": "your_token_here" }
    }
  }
}

Tip Claude Desktop does not inherit your shell PATH. If npx is not found, use the absolute path from which npx.

Quit Claude Desktop completely and reopen it.

claude.ai on the web

claude.ai runs connectors from Anthropic's cloud, not from your machine, so it needs a public HTTPS URL rather than a local command.

npx -y @thenavidm/vimeo-mcp-cli@latest --http --port 8000

Host that behind a public HTTPS URL, then in claude.ai go to Customize, Connectors, +, Add custom connector, and paste the URL. It binds to 127.0.0.1, which is all a tunnel or a proxy on the same machine needs. To accept connections from outside the machine, set VIMEO_HTTP_HOST=0.0.0.0 together with VIMEO_HTTP_TOKEN, a secret every request must send: without one, the server refuses to listen beyond this machine.

Cursor

Cursor reads .cursor/mcp.json, which takes the same JSON shape as Claude Desktop under the key mcpServers.

Windsurf

Windsurf reads ~/.codeium/windsurf/mcp_config.json under the key mcpServers.

VS Code

VS Code reads .vscode/mcp.json, where the key is servers rather than mcpServers, and each entry takes "type": "stdio".

Codex CLI

Codex CLI reads ~/.codex/config.toml:

[mcp_servers.vimeo]
command = "npx"
args = ["-y", "@thenavidm/vimeo-mcp-cli@latest"]

[mcp_servers.vimeo.env]
VIMEO_PAT = "your_token_here"

Gemini CLI

Gemini CLI reads ~/.gemini/settings.json under the key mcpServers.

Everything else

Any stdio MCP client takes the same three things: the command npx, the args, and the env block.

To disconnect, remove the server from your MCP client's config, then delete the token in your app's Authentication section at developer.vimeo.com. The second step matters: removing the config stops this server using it, revoking it stops anything using it.

5. Check it worked 🩺

VIMEO_PAT=your_token npx -y @thenavidm/vimeo-mcp-cli@latest doctor

It prints the account, the plan and the token's scopes, names any tool that a missing scope disables, and tests whether the plan has analytics. Both of the things that actually go wrong show up here. It exits 0 when everything works, 1 when a scope is missing or the token is rejected, and 10 when no token is set.

6. What it costs to have connected

Both surfaces are the same program with the same 43 tools. The difference is when the model pays for them. Measured in Claude Code:

Claude Code's tool search is on by default: it sends only the tool names and the server instructions, and loads a tool's full definition when the model reaches for it. An app that loads every tool up front pays the first line on every message, whether Vimeo comes up or not. With the skill added, Claude Code also lists its one-line description, about 210 tokens.

To spend less, turn the server off when you are not using it, which in Claude Code is the /mcp panel. VIMEO_READ_ONLY=1 takes the 26 write tools off the list, leaving 17. Or install the CLI and add the server on the days it earns its place.

Measured on 2026-10-05 against 1.1.1, with Claude Code 2.1.286 on Claude Opus 5.5 (one short prompt with and without the server connected, once with ENABLE_TOOL_SEARCH=false and once with the default, the difference read from the API's own usage figures; SKILL.md the same way) and Codex 0.159.3 on gpt-6.1-sol:

The task was "find the command that moves several videos into a folder, and the flags it requires". Every tool loaded costs less because each tool no longer repeats $schema, additionalProperties and an execution block. Over the CLI, every 1.1.1 run read the general help twice and the command list before the command's help, five requests that each carry the conversation so far, and every 2.0.0 run asked which and read the help: three. Tool search costs the same within the spread of each version's own rounds. SKILL.md costs 121 more because it lists which and exit code 1 and says how approval works over MCP. Other apps and models count tokens a little differently, and tool-list characters divided by four are not API usage.

7. Tools 🛠️

There are 43 tools, of which 17 are reads, and those are the only ones that remain under VIMEO_READ_ONLY=1.

Videos

Folders

Showcases

Chapters

Captions and transcripts

Comments

Account, stats and settings

8. Writing safely ✍️

Writes work by default. Organizing a library is the point of the tool.

Seven tools need confirming every time: the six deletes, and add_comment, because a comment is visible to everyone who can see the video the moment it posts. Nothing else is gated, because moving a video between folders or editing a title is one call to undo, and confirming everything teaches a model to confirm reflexively.

Over MCP a person approves each of those calls. Claude Code (2.1.246 and later) shows its own prompt, and a client that can show forms asks with an approval form whose one box starts unticked. Where a client can do neither, the model's confirm: true still counts, and VIMEO_CONFIRM=model makes it enough everywhere, for an agent with no person to ask. In a terminal the flag is --confirm, which --agent never adds.

Two flags deserve their own mention. delete_videos_too on delete_folder and remove_videos_from_folder destroys videos rather than unfiling them. Both default to false. delete_folder always needs confirming, and remove_videos_from_folder only when the flag is set.

remove_videos_from_folder tells clients the most it can do, destroy videos, even though unfiling them is a reversible write.

Comment text is written by other people and can contain instructions aimed at whatever reads it. Every comment comes back wrapped and labeled as viewer-authored data, which helps and is not a guarantee. For an agent working unattended on other people's content, VIMEO_READ_ONLY=1 is the real defence.

9. How Vimeo actually behaves 🎬

Scopes are frozen at creation. A token cannot gain a scope later. delete and video_files are both off by default, and Vimeo's 403 does not name the missing scope, so a failing delete looks like a missing video. doctor names it.

A 404 can mean "your plan does not include this". Analytics and teams both answer 404 rather than 402 on a free account. get_video_analytics translates that into a plain sentence instead of passing the confusion along.

Two things are called analytics. stats.plays is a lifetime counter on every plan. The analytics API is views over time and finish rate, on paid plans.

A video lives in exactly one folder. Adding it to a folder moves it. A showcase is different: any video can be in any number of showcases, and adding it there moves nothing.

Vimeo expands nested objects in full. Asking for a video's parent_folder returns the folder, its owner, that owner's avatar in nine sizes and their upload quota, about two kilobytes to learn a folder name. This server always requests specific leaf fields, which is why its output is small.

The bulk endpoints are inconsistent. Folder add and remove take the video list as a uris query parameter. The showcase bulk endpoint takes a videos body field and it replaces the showcase contents rather than appending, which is why this server does not expose it: adding to a showcase goes one video at a time so a curated list cannot be wiped by accident.

Transcripts are two requests. The API returns caption metadata with a link, and the text lives at that link on a CDN. The link is signed and short-lived, so it cannot be cached or handed to someone to open later.

Uploads are asynchronous. upload_video returns a video id straight away while transcoding continues. Watch status on get_video until it reads available.

10. This and Vimeo's own connector ⚖️

Vimeo publishes its own hosted MCP connector at mcp.vimeo.com. It is good, it is free with a Pro plan, and for a lot of people it is the right choice. Here is the honest split, so you can pick rather than guess.

Use Vimeo's if you want analytics or their AI features. It has viewer retention curves and retention insights, AI video summaries, semantic search inside a video, teams, staff picks, and an editing and render pipeline. None of that is reachable through the public API, so nothing here can match it.

Use this one if you want to change your library rather than read it.

They are not rivals and nothing stops you running both. Theirs answers what happened to a video. This one changes what the library looks like.

11. Your data 📂

There is no backend. This server runs on your machine and talks to api.vimeo.com directly.

It stores nothing: there is no session file, no cache and no database. The only file it ever writes is the audit log, and only when you set VIMEO_AUDIT_LOG to a path. That file holds, per attempted write, a timestamp, the tool, its risk, a one-line summary and what happened, who approved it when it needed approval, and a line when it finished. It is created with 0600 permissions.

Your token lives wherever you put it, which is your MCP client's config file. Nothing else is transmitted anywhere.

Settings

The program reads the environment directly. It does not load .env files.

Credentials

Safety

Tuning

12. Troubleshooting 🔧

Run doctor first. It catches most of this. Personal access tokens do not expire on a timer: they stop when revoked, or when their app is deleted.

13. FAQ

An MCP server is a standard way to give an AI assistant real access to a tool, so it can act rather than guess. You install it once, your assistant gains the tools, and it works in Claude, Cursor, ChatGPT and anything else that speaks MCP.

vimeo-cli is the same program as the MCP server, run as commands. AI agents that run commands, like Claude Code, Codex and OpenCode, use it on their own, and you can type the same commands in a terminal, a script or a cron job. Every tool is a command with dashes, so list_videos runs as vimeo-cli list-videos.

Use the MCP server in an app with no terminal, like Claude Desktop's chat. Use the CLI anywhere commands run: an agent like Claude Code, Codex or OpenCode, a script or a cron job. The MCP server's tools take up context on every message, and the CLI costs nothing until it runs.

Vimeo is a video hosting platform. People use it for work that needs to look professional and stay private: course content, client work, internal training, webinars. Unlike a public video site, you control exactly who can watch each video and which sites are allowed to embed it.

You need to paste a token into a config file once. The setup section has a prompt you can hand to an agent that will do the config part for you. After that you talk to it in plain language.

There is no backend and nothing is collected. The server runs on your machine and talks to Vimeo's API directly. Whatever you ask your AI assistant is governed by that assistant's own privacy terms, not by this server.

It does bulk work, mostly. Filing a hundred videos into folders is one call here and a hundred drags in the web app. Reading a transcript and turning it into chapter markers is a normal request here and manual work there. Everything it does is something the Vimeo API allows.

Deleting is guarded twice. Six tools need confirming before they run: delete_video, delete_folder, delete_showcase, delete_chapter, delete_texttrack and delete_comment. Over MCP a person approves each one in the client's own prompt or form; in a terminal it takes --confirm. On top of that, most tokens do not hold the delete scope, so those tools fail outright unless you deliberately ticked that box.

The one to watch is delete_videos_too, an option on delete_folder and remove_videos_from_folder that destroys videos instead of unfiling them. It defaults to false and needs confirming when set. If you want no risk at all, run with VIMEO_READ_ONLY=1.

They do different jobs. Vimeo's is hosted, needs a Pro plan, and is built for reading: viewer retention, AI summaries, moment search and their editing pipeline. It deliberately cannot delete anything, cannot upload, and cannot write to folders.

This one runs locally on any plan including free, and is built for changing a library: uploading, deleting, moving videos between folders in bulk, writing chapters and captions, managing tags and embed presets.

If you want to know how a video performed, use theirs. If you want to reorganize a library, use this. Running both is fine. Section 9 has the full table.

The server is free and MIT licensed. You need a Vimeo account, and Vimeo's own plan limits apply: analytics and teams need a paid plan, and upload limits depend on your tier.

It works with any client that speaks MCP. Configs for Claude Code, Claude Desktop, claude.ai, Cursor, Windsurf, VS Code, Codex CLI and Gemini CLI are all in section 4.

You cannot do it in one server instance, because a token belongs to one account. To use two, add the server twice under different names with a different token in each.

14. What changed 📋

Every release is in CHANGELOG.md, newest first.

Questions

Run into a problem or have a question? Open an issue and I will help.

Built by

Navid Moazzez

Navid Moazzez is a leading AI business strategist, and the host of the AI Creator Summit, watched by 100,000+ creators. He helps creators and founders master AI and build their own AI Operating System (AI OS) to automate their business and life. He creates useful free tools, MCP servers and CLIs that creators and founders can use in their own workflows.

Dependencies

License

MIT. Free to use, modify, and share.

Not affiliated with, endorsed by, or connected to Vimeo.com, Inc. Vimeo is a trademark of Vimeo.com, Inc.

© 2026 Navid Media. Made with ❤️ by Navid Moazzez.