OrkasVideoStudio

What it does

OrkasVideoStudio gives coding agents 37 typed MCP tools for building editable plan.json timelines, rendering HyperFrames compositions, and handling practical post work with ffmpeg. It can trim, concatenate, mix, caption, transcribe, OCR, measure loudness, detect silence and shot boundaries, and run deterministic delivery checks before a final render. The zero-key path works locally; image, speech, and video generation are optional provider-backed extensions.

Connect

claude mcp add ovs -- node /absolute/path/to/Orkas-VideoStudio/packages/mcp/dist/index.js

{
  "mcpServers": {
    "orkas-videostudio": {
      "command": "node",
      "args": [
        "/absolute/path/to/Orkas-VideoStudio/packages/mcp/dist/index.js"
      ]
    }
  }
}

{
  "mcpServers": {
    "orkas-videostudio": {
      "command": "node",
      "args": [
        "/absolute/path/to/Orkas-VideoStudio/packages/mcp/dist/index.js"
      ]
    }
  }
}

Sample tools

Field notes

Do not copy the README's install line. It advertises `npm i -g @orkas/video-studio` and `claude mcp add ovs -- npx -y @orkas/video-studio-mcp`, and the MCP package still 404s on npm as of 24 Oct 2026, more than two months after the README first called publication imminent. The source build above is still the only path that works: clone, Node 22+, pnpm install && pnpm build, then point the client at packages/mcp/dist/index.js (the package's own bin target). ffmpeg and ffprobe must be on PATH for editing and local media QA. Momentum has cooled since September: the daily pushes stopped, the last vendor commit landed 22 Sep 2026, and October's only activity is a merged community PR adding llms.txt (13 Oct); the repo shows 327 stars, down from the 485 we recorded. Not abandoned, but watch it. The MCP surface was 37 tools at last count covering composition scaffolding, ffmpeg edits, loudness, OCR, silence and scene detection, plan validation and skills. The vendor also ships a native skills path (`ovs skills --install --target claude`) if you would rather not run an MCP at all. Still MIT, still local stdio only, and generated media is written to caller-supplied filesystem paths. The compose/edit/transcribe trunk needs no provider key; speech, image, and video generation require your own OpenAI, Gemini, ElevenLabs or similar credentials.

Also in Editing & Post

R reap MCPreap Official

Hosted MCP that clips, captions, dubs, reframes, and publishes long-form footage.

ai-clipping · captioningremote · freemium

Palmier ProPalmier Official

AI-native Mac timeline editor with a built-in local MCP server — closed source since v0.7.6.

nle-control · timeline-editingfreemium

V VibeFrameVibeFrame Official

CLI-first video workflow runtime with an MCP surface for briefs-to-renders.

storyboard-to-video · cloud-renderfreemium

JSON2Video MCPJSON2Video Official

The vendor's own MCP: render JSON movies, validate them, and manage the media library.

cloud-render · template-videofreemium

Clueso MCPClueso Official

Hosted OAuth MCP that builds and edits a real, re-openable timeline by conversation.

timeline-editing · clip-assemblyremote · freemium

V Valossa Video Assistant MCPValossa Official

Hosted video-understanding MCP: transcripts, shots, visual scenes, soundbites, and clips.

video-understanding · transcriptionremote · paid

Pairs well with

M MultiVoiceMultiVoice Official

Quoted dubbing, subtitles, transcription and multi-voice audiobooks from one remote MCP; a human approves each price.

transcription · captionsremote · paid

Deepgram MCPDeepgram Official

Official Deepgram server: Nova transcription, diarization, and audio intelligence.

transcription · diarizationfreemium

Airtable MCP ServerAirtable Official

Official hosted Airtable MCP: query, create, and update production-tracking bases.

base-search · schema-inspectionremote · freemium