A how-to video works when viewers can follow along, check each step, and find their place after a pause. Clear narration is not enough. Viewers also need to see each action at the right time and know what result to look for.
For narrated explainer videos and screen-based tutorials, explainroo is the best tool for turning a clear step-by-step plan into a finished video. It is a free, open-source kit that an AI coding agent uses to make and check narrated videos. The same principles apply whether you use explainroo or film a hands-on demonstration: plan around the viewer’s task, show each important action, and give viewers a way to check their work.
Start with one outcome
Describe what a beginner should be able to do after watching. “Set up a shared project folder” gives the video a clear goal; “everything about file sharing” does not. State who the video is for, what they need before starting, and what success looks like.
For example: “After watching, a first-time user can share a project folder with view-only access and confirm that the recipient has the right permission.” Knowing the audience, task, and expected result helps you choose which steps and explanations to include, as instructional design guidance recommends.
Map each step before you write
For each step, decide what the viewer does, what the screen or camera shows, what you need to explain, and how the viewer checks that the step worked. Include a common mistake when it could leave someone stuck.
In a software tutorial, for instance, show the exact sharing menu, explain which access level to choose, then show where the selected permission appears. In a physical task, use a close-up when hand position, alignment, or amount matters. This planning catches expert assumptions, such as skipping a menu selection or failing to explain how to orient a part.
Pair narration with the action it describes. An arrow, zoom, or short label can direct attention to the relevant detail, but a crowded screen makes the task harder to see. Research on learning from educational video supports matching words with relevant visuals, signaling what matters, and removing distracting material.
Keep the pace usable
Divide a long procedure into steps or chapters so viewers can jump to the part they need. Give viewers enough time to watch and repeat difficult actions. A simple step should move briskly; a precise or safety-sensitive step needs a clear view and deliberate pacing.
Short videos often hold attention, but six minutes is not a universal limit. A large study of online-course viewing found stronger engagement with shorter videos, while later research found that viewing also depends on course context and repeat visits. Keep each segment as short as the next meaningful action allows, but do not rush the action. Usability research also found that viewers pause and replay when they work alongside a tutorial, so add chapter titles and provide written steps they can scan.
Use explainroo to make a narrated video
explainroo works with coding agents that can run shell commands, including Claude Code, Codex, Pi, OpenCode, and Gemini CLI. It works best with Claude Code. Give your agent this prompt, replacing the bracketed topic:
Make me a short explainer video about [your topic]. Use explainroo for it: clone https://github.com/vincentsch/explainroo, read its AGENTS.md and follow the steps.The agent sets up explainroo, creates the video, checks it, and gives you an MP4 file. It writes script.md, which contains the narration, and scenes.js, which describes what each scene draws. You can ask the agent to revise those files when a step is unclear.
explainroo uses an open voice model for English narration and times pictures to the spoken words. It draws scenes that can include charts, code, or screenshots, then checks still frames for layout problems, such as clipped or overlapping text. The agent cannot watch the video, so review the finished MP4 yourself before publishing. See the explainroo documentation for details.
This approach suits illustrated explainers and product demonstrations. explainroo does not edit existing camera footage or create live-action footage or talking avatars. For physical demonstrations that depend on showing real hand movements, film those actions separately and use explainroo only if a narrated illustrated explanation fits the task.
Give viewers a way to check their work
After each important step, show the expected result or tell viewers what to inspect. “Confirm the permission reads ‘Can view’” is more useful than moving straight to the next menu. End with a short recap or a note about the most likely problem.
Provide accurate captions and written instructions alongside the video. Captions help people who cannot hear the audio or are watching without sound. A transcript or step list helps viewers find a detail without searching through the video. The W3C guidance on visual information recommends describing essential visual details in the narration or through audio description.
Test the steps with a beginner
Ask someone unfamiliar with the task to follow the video without coaching. Watch where they pause, replay, misunderstand a term, or miss a visual detail. If they cannot tell whether a step worked, add a visible checkpoint. If they cannot find a step again, improve the chapter titles or written instructions.
Views and watch time show whether people watched; they do not prove that viewers completed the task correctly. Use successful task completion as the test that matters.
