Showhow Explainer Reels: Free Code-Drawn Video in 2026

Showhow explainer reels are short videos where the visuals are drawn in code and every animation is timed to land on a spoken word. Showhow is a free, open-source project from tein, published at github.com/sirioberati/showhow. It draws each move in code, syncs it to narration, and checks every render before you publish. You run it locally and drive it from an AI coding agent such as Claude Code.

narration script
      |
      v
+------------------+     +-------------------+
| split into words |---->| time each word    |
+------------------+     +-------------------+
                               |
                               v
+------------------+     +-------------------+
| code draws moves |---->| render the reel   |
+------------------+     +-------------------+
                               |
                               v
                      +-------------------+
                      | check the render  |
                      +-------------------+
                               |
                               v
                    publish to social accounts

What Is Showhow and How Does Word-Synced Code Animation Work?

Showhow turns a narration script into an explainer reel where the drawing is code, not a timeline of clips you drag by hand.

The core idea is simple. You give it a script. It works out when each word is spoken. Then it draws shapes, text, arrows, counters, diagrams and camera moves in code, and schedules each of those moves to fire on the word it belongs to. When the narrator says “three times faster”, the number three appears on that word, not two frames later.

This matters because most explainer reels fail on timing. A caption or a chart arrives late, the viewer’s eye drifts, and the point is lost. Code-drawn animation removes the manual keyframing step. You describe the move once, in code, and the tool places it against the audio.

Showhow also checks each render. That means it looks at the output before calling it done, which is the part most local video scripts skip. A checked render is the difference between a file you still have to scrub through and a file you can post.

Get started for $0

How Showhow Times Every Move to a Spoken Word

The timing model is the whole product: every move is bound to a word, so the animation and the voice cannot drift apart.

Here is the shape of the workflow in plain steps:

1. Write the narration as text. Keep sentences short and put one idea per sentence.
2. Showhow splits the narration into words and assigns each word a time.
3. You attach moves to words in the script. A move can be a shape appearing, a line drawing itself, a number counting up, or text sliding in.
4. Showhow renders the reel with those moves placed on their words.
5. The render is checked before it is handed back to you.

A worked example. Take the line: “Your API key stays in one workspace.” You draw a small box on the word “API”, a key icon on “key”, and a fence around the box on “workspace”. Each of those is a code move with a word anchor. If you later re-record the narration a second faster, the moves follow the new word times instead of breaking.

That is the practical gain. Re-timing a hand-edited reel after a voice change is an afternoon. Re-timing a word-anchored reel is a re-run.

Where to Get Showhow: GitHub Repo and Setup

Showhow lives at github.com/sirioberati/showhow, and setup is a clone plus a local run, driven from your AI coding agent.

The repo page is the source of truth for install commands, dependencies and the render pipeline. Read it before you start, because a code-drawn video tool depends on the local tools on your machine: a language runtime, a renderer, and audio input for the narration.

Two habits make the first run easier:

• Start with a 20 second script, one idea, one move per sentence. Get a finished render before you attempt a 90 second explainer.
• Keep the narration text in its own file and treat it as the source. The moves point at words in that file, so editing the file is how you retime the reel.

Because Showhow is driven by an agent rather than a GUI, the setup you are really doing is teaching the agent the pipeline. Once it knows the repo, it can write the moves, run the render and read the check result without you touching the terminal.

Is Showhow Completely Free and Open Source?

Yes. The project describes itself as free and open source, and the source is public on GitHub, so you can read how the word timing and the render checks work before you trust them.

“Free” here means the software costs nothing. It does not mean zero cost overall:

Open source also means open licence terms. Check the licence in the repo for what you can do with the code in your own products. That is a normal step, not a warning.

Does Showhow Run on Free VRAM and Local Hardware?

Showhow draws in code and renders locally, and the project’s own repo is the place to confirm what hardware it needs.

The honest answer: VRAM requirements depend on the renderer and any model the pipeline uses, and that is documented in the repo, not something to guess at. What you can plan for:

• Code-drawn visuals are light. Shapes, text and diagrams do not need a large model to exist; they need a renderer.
• Audio work is the part that can want a GPU, depending on how the narration and word times are produced.
• Render length scales with resolution and duration. Short reels are cheaper than long ones in wall clock time, not just disk.

If you are on a laptop with no discrete GPU, start with the shortest possible script and measure a full render. If it completes in a reasonable time, you are fine for explainer reels. If the check step is slow, cut resolution before you cut the check, because the check is the feature.

Can You Use Showhow From Claude Code?

Yes. Showhow fits the Claude Code workflow, which is how most people will drive it: you describe the reel, the agent writes the code moves, runs the render and reads the result.

The loop looks like this inside a Claude Code session:

You: read the narration in script.txt and build a 30 second
     explainer reel. Put one move on every key noun. Then
     render it and tell me what the check reported.

Agent: writes the moves, runs the render, reads the output,
       reports back with any warnings and the file path.

This is the same pattern Creator OS uses for the rest of the pipeline. Creator OS ships a hosted MCP server at https://mcp.creatoros.ca/mcp and a CLI, so the agent that builds the reel is also the agent that publishes it. There are about 160 MCP tools, including create_post, upload_media_from_url, get_post_analytics and get_best_time_to_post.

The Claude Code path is one line to add the server:

claude mcp add --transport http creatoros \
  https://mcp.creatoros.ca/mcp \
  --header "Authorization: Bearer cos_live_..."

If you would rather install a plugin, the Creator OS plugin adds the MCP server, agent skills, a live pane and a publish guard that asks before anything is posted or spent:

/plugin marketplace add kevinbadi/creatoros-claude-plugin
/plugin install creatoros@creatoros
/plugin configure creatoros@creatoros

Watch the walkthroughs on the KevBuildsApps YouTube channel and the launch video to see the agent running the whole loop.

How Showhow Fits AI-Assisted Video Production for Social

Showhow covers one stage of the pipeline well, the part where a script becomes a timed, checked reel, and hands the finished file to whatever publishes it.

The stage split that works:

A concrete run from the terminal once the file exists:

creatoros media:upload reel.mp4
creatoros posts:create --text "How word-synced explainer
  reels work" --platforms instagram,tiktok,youtube \
  --media med_123 --scheduledAt 2026-07-14T09:00:00Z \
  --timezone America/Toronto

Instagram Trial Reels are worth knowing here. Post with post_type short_video and trial_reel set to "auto" or "manual", and Instagram shows the reel to non-followers first. Instagram only allows Trial Reels on accounts with 1,000+ followers, and Creator OS refuses the post up front below that with the error trial_reel_needs_followers.

After it goes live, an agent can read Reel watch stats per post through get_post_analytics: reelsSkipRate (the share of views that swiped away in the first 3 seconds), igReelsAvgWatchTime and igReelsVideoViewTotalTime, in milliseconds. Meta reports these for Reels only. That is the feedback loop: the number tells the agent whether the first move landed, and the next script changes. Creator OS provides the posting and the numbers; your prompt or workflow decides what to change.

If you want an agent to run the account on its own, Social Agents is the open-source harness for it. Start with npm start creatoros social-agents, answer the brand interview, and it posts, replies and reports. The full detail is in the social agents docs and the code-drawn explainer reel post.

Showhow vs Other AI Explainer Reel Generators

Showhow takes a different route from a browser-based explainer video builder: the visuals are code, the timing is word-anchored, and it runs on your machine.

That trade looks like this in practice:

Creator OS is not a video renderer. It is the publishing and analytics layer: 11 platforms plus Skool, the API, the CLI, and a hosted MCP server so an agent can post, reply to comments, read analytics and run automations. Showhow makes the reel; Creator OS ships it and reports on it. Both are pieces of the same AI-assisted workflow, and the pairing is the point. The animated explainer walkthrough covers the model side of that pairing.

FAQ

Is Showhow completely free to use?

The software is free and open source, so there is no licence fee to clone the repo and render reels. Your real costs are local hardware, time, and however you produce the narration audio. Check the licence in the repo for how you can reuse the code.

Apakah Showhow gratis dan open source?

Ya, Showhow gratis dan open source. Kode sumbernya tersedia di GitHub, jadi kamu bisa membaca cara kerja penyesuaian kata dan pengecekan render sebelum memakainya. Biaya yang tersisa adalah perangkat lokal dan produksi narasi, bukan lisensi perangkat lunak.

How does Showhow sync animations to spoken words?

It splits the narration into words and gives each word a time. You anchor each move to a word in the script, so the move fires on that word during the render. Re-record the narration and the moves follow the new timings instead of drifting.

Can I run Showhow from Claude Code?

Yes. Claude Code is a natural driver for Showhow: you describe the reel, the agent writes the code moves, runs the render and reads the check result. Add the Creator OS MCP server in the same session and the agent can publish the finished file and read the analytics afterward.

What are the hardware or VRAM requirements for Showhow?

That depends on the renderer and any model the pipeline uses, and the repo is where those requirements are stated. Code-drawn shapes and text are light; audio work is the part that can want a GPU. Start with a short script and time a full render on your own machine before committing to longer reels.

Get started

Clone Showhow, write a 20 second script with one move per sentence, and let your agent render it. Then hand the file to Creator OS and schedule it. Plans start at $14.99/month for 11 platforms plus Skool, with the API, the MCP server and the CLI included. Start your free trial at creatoros.ca/sign-up, and if you want the agent-first setup, read the MCP docs or the explainer video content calendar post for the posting side. More walkthroughs on KevBuildsApps.

Get started for $0
Free trial for new subscribers. Cancel anytime.