manim-skill MCP — Anwendungsfälle, Installieren & Live-Demo

Warum nutzen

Hauptfunktionen

Autonomous plan → write → render → refine loop
Scene library patterns for common math / CS visuals
Invokes local manim install (CE)
Handles low-quality preview vs high-quality final renders
Iterates on feedback ('make the graph animate left-to-right')

Live-Demo

In der Praxis

manim-skill.replay ▶ bereit

0/0

Installieren

Wählen Sie Ihren Client

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

{
  "mcpServers": {
    "manim-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Yusuke710/manim-skill",
        "~/.claude/skills/manim-skill"
      ],
      "_inferred": true
    }
  }
}

Öffne Claude Desktop → Settings → Developer → Edit Config. Nach dem Speichern neu starten.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "manim-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Yusuke710/manim-skill",
        "~/.claude/skills/manim-skill"
      ],
      "_inferred": true
    }
  }
}

Cursor nutzt das gleiche mcpServers-Schema wie Claude Desktop. Projektkonfiguration schlägt die globale.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "manim-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Yusuke710/manim-skill",
        "~/.claude/skills/manim-skill"
      ],
      "_inferred": true
    }
  }
}

Klicken Sie auf das MCP-Servers-Symbol in der Cline-Seitenleiste, dann "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "manim-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Yusuke710/manim-skill",
        "~/.claude/skills/manim-skill"
      ],
      "_inferred": true
    }
  }
}

Gleiche Struktur wie Claude Desktop. Windsurf neu starten zum Übernehmen.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "manim-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Yusuke710/manim-skill",
        "~/.claude/skills/manim-skill"
      ]
    }
  ]
}

Continue nutzt ein Array von Serverobjekten statt einer Map.

~/.config/zed/settings.json

{
  "context_servers": {
    "manim-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/Yusuke710/manim-skill",
          "~/.claude/skills/manim-skill"
        ]
      }
    }
  }
}

In context_servers hinzufügen. Zed lädt beim Speichern neu.

claude mcp add manim-skill -- git clone https://github.com/Yusuke710/manim-skill ~/.claude/skills/manim-skill

Einzeiler. Prüfen mit claude mcp list. Entfernen mit claude mcp remove.

Anwendungsfälle

Praxisnahe Nutzung: manim-skill

Animate the intuition behind a math or CS concept

👤 Educators, YouTubers, students explaining to themselves ⏱ ~45 min intermediate

Wann einsetzen: A concept is easier with motion than with words (gradient descent, sorting, Fourier).

Voraussetzungen

manim installed — pip install manim (or uv add manim); ensure ffmpeg present
Skill installed — git clone https://github.com/Yusuke710/manim-skill ~/.claude/skills/manim-skill

Ablauf

State the concept

Use manim-skill. Animate gradient descent on a 2D loss surface — ball rolling, learning rate visibly affects step size.✓ Kopiert

→ Scene plan + code + low-q preview rendered
Iterate

Slower at first, faster once it's in the valley. Label the axes.✓ Kopiert

→ Code updated, re-rendered preview
Final render

Looks good — render at 1080p60.✓ Kopiert

→ High-q mp4 produced

Ergebnis: An explainer video you can embed in a blog post.

Fallstricke

Manim code has breaking changes across versions — Pin your manim version in requirements; the skill's prompts assume recent Community Edition

Kombinieren mit: filesystem

Build a series of scenes for a longer explainer

👤 Video creators working on a single concept over multiple shots ⏱ ~180 min advanced

Wann einsetzen: You have a 5-minute script and need 10+ scenes to go with it.

Ablauf

Feed the script

Use manim-skill. Here's my script. Break into scene beats and propose a Manim scene per beat.✓ Kopiert

→ Ordered scene plan
Render in batches

Render scenes 1-3 at preview quality; show me.✓ Kopiert

→ Three low-q mp4s
Consolidate

Ok, lock scenes 1-3. Move to 4-6.✓ Kopiert

→ Work proceeds; locked scenes untouched

Ergebnis: A full scene library ready for final render.

Fallstricke

Rendering high-q for every iteration wastes time — Iterate at -ql (low) and only go -qh once locked

Kombinationen

Mit anderen MCPs für 10-fache Wirkung

manim-skill + filesystem

Organize a project's scenes, renders, and drafts

Keep scenes in project/scenes/, renders in project/renders/preview or /final.✓ Kopiert

Werkzeuge

Was dieses MCP bereitstellt

Werkzeug	Eingaben	Wann aufrufen	Kosten
plan_scenes	concept or script	First step on any new project	0
write_scene	scene spec	Per-scene code generation	0
render_manim	scene file, quality	Preview and final renders	local CPU/GPU
refine_scene	scene + feedback	Iterating on animation	0

Kosten & Limits

Was der Betrieb kostet

API-Kontingent: None
Tokens pro Aufruf: Moderate — code iterations
Kosten in €: Free; local compute only
Tipp: Always iterate at low quality. Final render once.

Sicherheit

Rechte, Secrets, Reichweite

Credential-Speicherung: None

Datenabfluss: None from skill

Fehlerbehebung

Häufige Fehler und Lösungen

manim import errors

Manim CE vs GL — make sure you're on Community (pip install manim), not the legacy 3b1b version

Prüfen: python -c 'import manim; print(manim.__version__)'

Render hangs or is very slow

Use -ql for iteration; -qh only at end. Check ffmpeg presence.

Prüfen: ffmpeg -version

Weird layout — objects overlap

Use positioning primitives (next_to, align_to) rather than raw coords

Alternativen

manim-skill vs. andere

Alternative	Wann stattdessen	Kompromiss
After Effects / Motion Canvas	You want visual timeline tools	Not code; not LLM-friendly
paperbanana-skill	Static academic diagrams rather than animations	Different output

Mehr

Ressourcen

📖 Offizielle README auf GitHub lesen

🐙 Offene Issues ansehen

🔍 Alle 400+ MCP-Server und Skills durchsuchen