AI Explainer Video Generator
Give Claude a topic, or nothing at all. It researches, writes a fact-checked script, generates editorial collage or paper-diorama clips, adds a narrator and subtitles, and hands you one finished video.
Set up in ClaudeAbout 441 credits per finished minute · 141 on the budget model
Why one island makes the world's chips
The ship that stopped world trade
Both made end to end by the skill: script, clips, voice and subtitles. Sound on.
From one sentence to a finished video
The agent runs every step on its own and only stops to show you the plan and credit estimate.
Research
Picks a trending topic if you give none, reads sources and cross-checks every number.
Script
Writes 10-second blocks: cold open, stakes, evidence, the turn and a closing line.
Style key
Locks the look with one reference image attached to every clip, so scenes match.
Clips
One AI clip per block that illustrates its own line, with no garbled on-screen text.
Voiceover
One narrator for the whole video. Each take is measured to fit its clip.
Assembly
Stitches clips and takes in sync, burns bold subtitles and returns one mp4.
Two house styles
Editorial collage
Archival photo cutouts, torn paper, tape, marker circles and growing paper charts on flat mustard, navy and coral. Best for data stories, tech, money and "why does X" questions.
Paper diorama documentary
A handmade miniature world of aged newspaper and cardboard, anonymous cutout figures and one burnt-orange object, lit like a film set. Best for geopolitics, history and power.
Want another look? Describe it, and the agent generates a new style key before the first clip.
Two minutes, once
The imageat MCP server provides the tools; the skill teaches the agent the pipeline.
- 1
Connect imageat
In Claude, open Settings → Connectors → Add custom connector, paste the URL below and sign in with your imageat account. Step-by-step guide.
Connector URLhttps://mcp.imageat.com/mcp
- 2
Upload the skill
Download imageat-explainer-video.skill and upload it under Settings → Capabilities → Skills.
- 3
Ask for a video
Try “Make a 1-minute explainer video about why concert tickets got so expensive.” The agent posts a plan with a credit estimate, then delivers the mp4, the script and its sources.
Things to ask for
What a minute costs
Default · Gemini Omni 720p
~441credits / min
Best motion and style adherence.
Budget · Wan 3.0 480p
~141credits / min
Ask for “the cheap version” or name wan3.
Includes voiceover (18 credits per minute) and assembly with subtitles (3 per minute). Failed clips are refunded. Plans & credits
imageat-explainer-video.skill
One file that teaches Claude the whole explainer workflow on top of the imageat tools. Upload it once in Settings → Capabilities → Skills.
- The full six-step pipeline, from topic research to the assembled mp4
- Script rules: block length, structure and a narrator-friendly rhythm
- Prompt templates and vocabulary for the collage and diorama styles
- Hosted 9:16 and 16:9 style keys, so every clip matches
- Voice timing checks, cost estimates and what to do when a step fails
Want images and single videos too? See the imageat skill and the MCP server. Prefer the web app?
FAQ
What does the AI explainer video generator make?
A finished narrated video: a researched, fact-checked script, one AI clip per 10-second block in a consistent style, a documentary voiceover, burned-in subtitles and a single assembled mp4. Vertical 9:16 for Shorts, TikTok and Reels by default, or 16:9 for YouTube.
How do I use it?
Connect the imageat MCP server to Claude (or another MCP client), upload the imageat-explainer-video.skill file, and ask for a video about any topic. The agent researches, writes, generates and assembles the video on its own and sends you the mp4 with the script and sources.
How much does a video cost?
About 441 credits per minute with the default Gemini Omni clips, or about 141 credits per minute with Wan 3.0 at 480p. Voiceover is 3 credits per block and assembly with subtitles is 3 credits per minute. The agent posts an estimate before it starts.
Is this a Vox-style video generator?
It makes the editorial collage look that explainer channels such as Vox made popular: cutouts, paper textures, marker annotations and animated charts, with a calm narrator. imageat is not affiliated with Vox Media; the videos, scripts and voices are generated for you.
Can I pick the voice, language or length?
Yes. Ask for any of the 40 voices (or let the agent pick a documentary narrator), another narration language, a length from 30 seconds to 10 minutes, landscape framing, or no subtitles. Anything you don't specify uses sensible defaults.
What if a voiceover runs longer than its clip?
The assembly step measures every clip and take before charging. If a line runs long it rejects just that block with a clear message, the agent shortens and re-voices it, and assembles again. Failed clips and failed assemblies are refunded automatically.
Can I use the videos commercially?
Yes, the videos you generate are yours to publish, including on monetized channels. Check facts and sources before publishing, as you would with any script.