claude-shorts MCP — Use Cases, Install & Live Demo

Why use it

Key features

AI-scored segment extraction from long-form content
Audio-aware boundary snapping so clips don't cut mid-word
Remotion-rendered animated captions
Cursor tracking overlays for screen-capture sources
Python orchestration end-to-end

Live Demo

What it looks like in practice

claude-shorts-skill.replay ▶ ready

0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "claude-shorts-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/AgriciDaniel/claude-shorts",
        "~/.claude/skills/claude-shorts"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "claude-shorts-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/AgriciDaniel/claude-shorts",
        "~/.claude/skills/claude-shorts"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "claude-shorts-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/AgriciDaniel/claude-shorts",
        "~/.claude/skills/claude-shorts"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "claude-shorts-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/AgriciDaniel/claude-shorts",
        "~/.claude/skills/claude-shorts"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "claude-shorts-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/AgriciDaniel/claude-shorts",
        "~/.claude/skills/claude-shorts"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "claude-shorts-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/AgriciDaniel/claude-shorts",
          "~/.claude/skills/claude-shorts"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add claude-shorts-skill -- git clone https://github.com/AgriciDaniel/claude-shorts ~/.claude/skills/claude-shorts

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use claude-shorts

How to turn a 90-minute podcast into 10 Shorts

👤 Podcasters and content creators ⏱ ~90 min advanced

When to use: You published a long episode and want a week of Shorts from it with minimal effort.

Prerequisites

Node 20+ for Remotion and FFmpeg installed — nvm install 20; brew install ffmpeg
Skill cloned — git clone https://github.com/AgriciDaniel/claude-shorts ~/.claude/skills/claude-shorts

Flow

Transcribe and score

Process podcast_ep42.mp3 — transcribe and score candidate Short segments.✓ Copied

→ Ranked list of 20–30 candidates with timestamps and hooks
Pick and snap

Pick the top 10 and snap boundaries to clean audio breakpoints.✓ Copied

→ 10 segments with cleaned-up start/end
Render Shorts

Render vertical 9:16 Shorts with animated captions to ./shorts/.✓ Copied

→ MP4 files ready to upload

Outcome: A week's worth of Shorts generated in an afternoon.

Pitfalls

Captions out of sync — Use word-level timestamps from whisper — skill expects them
Same joke across 3 Shorts — Ask Claude to de-dupe by topic cluster before rendering

Combine with: filesystem

Build a highlight reel from a Twitch / YouTube stream

👤 Streamers, educators, conference archivists ⏱ ~60 min intermediate

When to use: You have a long recording and need a sampler to promote it.

Flow

Score peaks

Find the 15 most engaging moments based on transcript + audio energy.✓ Copied

→ Ranked peaks with timestamps and reasons
Assemble the reel

Assemble a single 60-second highlight reel with smooth cuts.✓ Copied

→ One MP4 with good pacing

Outcome: A concise reel that represents the best of the session.

Pitfalls

Reel has no narrative — Ask Claude to order clips for arc, not just score

Combine with: claude-youtube-skill

Combinations

Pair with other MCPs for X10 leverage

claude-shorts-skill + claude-youtube-skill

Same pipeline feeds YouTube long-form + Shorts strategy

Pick 10 Shorts from the latest podcast and suggest YouTube Shorts titles and descriptions.✓ Copied

claude-shorts-skill + social-media-caption-generator-claude-skill

Auto-caption each Short for IG/TikTok/Threads

For each rendered Short, generate posting captions for Instagram Reels and TikTok.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
transcribe-and-score	audio/video path	First step in any pipeline	Whisper / scoring compute
boundary-snap	segment timestamps + waveform	After scoring	audio analysis compute
remotion-render	segments + captions + style	Final step	Remotion render time
cursor-overlay	screen-capture + cursor events	For tutorials or software walkthroughs	0

Cost & Limits

What this costs to run

API quota: none beyond whisper/LLM provider
Tokens per call: ~10–30k tokens per long-form source
Monetary: free at the skill level; rendering uses your CPU/GPU
Tip: Score in one pass, then batch-render overnight to avoid UI blocking

Security

Permissions, secrets, blast radius

Credential storage: none

Data egress: none — local rendering

Respect the content licensing of the source you're clipping

Troubleshooting

Common errors and fixes

Remotion render fails

Check Node version (20+), ensure the Remotion project compiles: npx remotion render.

Verify: npx remotion --version

FFmpeg: codec not found

Install a full FFmpeg build: brew install ffmpeg; confirm libx264.

Verify: ffmpeg -codecs | grep 264

Captions off by ~200ms

Re-run transcription with word timestamps enabled.

Alternatives

claude-shorts vs others

Alternative	When to use it instead	Tradeoff
claude-youtube-skill	You need YouTube strategy, not automated rendering	Strategy vs pipeline

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills