Manim Animator logo

Manim Animator

Organization
sandbaseai
manim-animator

Create code-driven explanatory animations and precise vector graphics with Manim, optionally using SandBase for generated images, narration, music, or transcription. Use for mathematical, scientific, algorithmic, or diagrammatic animation; use a video-generation workflow for photoreal footage.

Overview

Publishersandbaseai
Repositorysandbase-skills
Skill namemanim-animator
Stars
200
Forks
21
Bundled files
1
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by sandbaseai on GitHub. Read the source before you install it.

Installation

Install the Manim Animator AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/sandbaseai/sandbase-skills/tree/main/marketing/manim-animator
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/sandbaseai/sandbase-skills.git /tmp/sandbase-skills
mkdir -p .claude/skills
cp -r /tmp/sandbase-skills/marketing/manim-animator .claude/skills/manim-animator
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Manim Animator in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Manim Animator on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Manim Animator is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Manim Animator

Turn an explanation into reproducible Manim source code and a verified video. Draw equations, diagrams, charts, and transitions locally; use SandBase only for generative media that Manim cannot produce itself.

Read the SandBase API map before generating any image or audio asset.

Route the request

Use Manim when the value comes from exact geometry, mathematical notation, deterministic timing, reproducible diagrams, or a step-by-step conceptual explanation. Prefer another workflow when the main deliverable is:

  • photoreal or cinematic generated footage;
  • conventional timeline editing of supplied footage;
  • a static chart or diagram with no meaningful animation;
  • an avatar or talking-head presentation.

Translate named stylistic references into concrete traits such as dark background, minimal palette, hand-drawn-feeling curves, spatial transformations, or deliberate pacing. Do not promise an exact imitation of a creator's identity.

Preflight

Inspect the current project before creating files:

  • determine whether it already uses Manim Community or ManimGL and do not mix their APIs;
  • check the installed Python and Manim versions, renderer, FFmpeg, and LaTeX availability;
  • preserve the project's existing environment and dependency manager;
  • identify output resolution, frame rate, aspect ratio, duration, codec, and caption requirements.

Default to Manim Community for a new project unless the request specifically depends on ManimGL behavior. Do not install system packages, change global environments, or add optional plugins without user authorization. If required tooling is unavailable, return the source and an exact dependency plan instead of claiming a render succeeded.

Plan scenes

For more than one short idea, create a compact scene plan before coding:

FieldRecord
PurposeQuestion or idea the scene resolves
DurationTarget seconds, or measured narration duration
ObjectsEquations, labels, shapes, axes, graphs, images, and camera elements
State changesWhat appears, transforms, moves, highlights, or disappears
ExplanationNarration or on-screen text associated with each beat
ContinuityObjects and visual state carried into the next scene
Technical needs2D/3D scene type, LaTeX, updater, plugin, raster asset, or audio

Keep each scene focused on one conceptual transition. Reuse object instances when continuity matters, and design equations so corresponding terms can transform cleanly.

Generate optional media through SandBase

Manim code and vector graphics are local work and do not require SandBase. For raster images, narration, music, sound effects, or transcription:

  1. Call sandbase_discover separately for the needed capability, such as text to image, text to speech, text to music, or speech to text.
  2. Call sandbase_inspect for serious candidates. Confirm language, timing controls, formats, reference-input rules, limits, price, and execution mode.
  3. Prefer vector construction over generated raster media when Manim can express the visual precisely.
  4. Before any paid call, show the selected endpoint, important arguments, call count, per-call price, and total estimate or uncertainty. Obtain confirmation for the exact call or batch.
  5. Use sandbase_account before an approved multi-call batch.
  6. Call sandbase_run with the exact discovered name and only current schema-defined arguments.
  7. Poll an asynchronous result with sandbase_run_get using the same run ID. Never submit a replacement merely because it is pending.
  8. Use sandbase_runs only to recover a result or reconcile actual cost.

Generate and save every asset before writing code that references its path. Keep an asset ledger with purpose, file path, SandBase endpoint, run ID, license or provenance, and actual cost when available.

Narration-first timing

When narration controls the pace:

  1. Finalize the narration text per scene.
  2. Generate one approved audio clip per scene rather than one monolithic file.
  3. Measure each clip with an available local media inspector.
  4. Use the measured duration as the scene timing budget.
  5. Make the sum of animation run_time and waits match that budget through meaningful pacing, not black-frame padding.

Use an inspected transcription capability only when captions or word-level timing are required and the selected schema provides suitable timestamps. Otherwise use scene-level caption timing derived from the narration plan.

Write maintainable Manim code

  • Keep configuration outside scene bodies when the installed Manim version supports project configuration.
  • Use named constants for palette, typography, margins, timing, and repeated coordinates.
  • Group related objects and use relative placement methods instead of unexplained pixel-like offsets.
  • Create axes, labels, and equations with explicit ranges and semantic names.
  • Prefer transforms between related states over delete-and-recreate sequences.
  • Keep updaters bounded, remove them when no longer needed, and avoid frame-dependent side effects.
  • Use camera movement only when it clarifies spatial relationships.
  • Ensure text contrast, minimum readable size, safe margins, and enough dwell time.
  • Avoid runtime network requests. Resolve external assets before rendering.

For multi-scene output, keep scenes independently renderable and maintain a deterministic source order for final assembly.

Render and iterate

  1. Render every scene at low quality first with caching disabled when stale output is possible.
  2. Capture the exit status and a concise error tail per scene.
  3. Fix import, API-version, LaTeX, missing-asset, geometry, and timing errors before visual review.
  4. Inspect representative frames at the beginning, key transformations, dense layouts, scene boundaries, and ending.
  5. Revise overlaps, clipping, illegible text, jerky transforms, excessive motion, and continuity breaks.
  6. Stop after three unsuccessful revision loops and return the source, logs, and remaining blocker.
  7. After all scenes pass, render at the requested final quality and concatenate or mix audio with local deterministic tooling.

Do not start a costly final render until a low-quality preview has passed technical and visual checks. Do not edit encoded frames by hand to hide a source-level defect.

Final verification

Check the final container, video and audio codecs, duration, dimensions, frame rate, decodability, audio presence, caption presence, and representative frames. Watch the complete render when the host supports playback. For narrated work, compare each scene's visual duration with its measured clip and check that speech is not cut off.

Return the Manim source, dependency or run command, final video, scene plan, generated-asset ledger with SandBase run IDs and costs, and any known rendering or timing compromises. Do not claim completion if a scene failed, a required asset is missing, or the final file was not checked.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Manim Animator AI skill do?

Create code-driven explanatory animations and precise vector graphics with Manim, optionally using SandBase for generated images, narration, music, or transcription. Use for mathematical, scientific, algorithmic, or diagrammatic animation; use a video-generation workflow for photoreal footage.

Why use Manim Animator on TypingMind?

Because you install it once and use it with any model. Manim Animator is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Manim Animator in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/sandbaseai/sandbase-skills/tree/main/marketing/manim-animator. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Manim Animator?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Manim Animator?

As many as you like. As long as a model supports skills, you can use Manim Animator with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Manim Animator AI skill free?

Yes. It is published on GitHub by sandbaseai under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇