Stickman Video Director Turns Copy Into Six Gemini Omni Flash Prompts
Turn copy into rich one-minute Gemini Omni Flash stickman video prompt packages.
At a glance
- What is it?
- A Codex Skill that reshapes your notes into a confirmed six-act director's proposal and six production-ready prompts for a one-minute stickman video. The confirmation gate is the interesting part; the audio consistency caveat is the honest one.
- Who is it for?
- Adopt it if you already write in Codex and want the story structure settled before you spend generation credits; the confirmation gate is the whole product. Skip it if you need a rendered video straight out of the tool, if you work outside Codex, or if you need frame-accurate audio continuity across the six clips, since the README admits independently generated segments can differ in voice and music.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 31 days ago.
- What is it written in?
- Mainly Shell, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 16, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The gap between having copy and having a one-minute video
A good idea does not guarantee a watchable minute. The README frames the failure mode directly: one character, one background, and almost no new visual change across ten seconds. Directing a full minute means designing an opening hook, pacing the explanation, inventing visual metaphors that fit the subject, moving the camera, linking scenes, and holding consistency across separate generations.
Stickman Video Director is aimed at that gap. It is a Codex Skill, so it runs inside a Codex conversation rather than as a standalone app. The intended user is someone producing explainer, motivational, or educational shorts for YouTube Shorts, TikTok, Instagram Reels, and YouTube, who already has raw material (copy, notes, an article, or just a topic) and wants the production thinking done before generation credits are spent. The README states there is no API and no MCP dependency; the whole workflow happens in the dialogue.
Paste, select, confirm, generate, stitch
The workflow has five named steps. You paste the copy. You select an aspect ratio (`16:9`, `9:16`, or `1:1`) and a light or dark theme. You confirm a director's proposal that contains the English voice-over, a reference translation, visuals, camera, transitions, BGM, and sound effects. Only after approval does the Skill emit six separate Gemini Omni Flash prompts. You then stitch six roughly ten-second clips into a coherent minute.
The mechanism that matters is the confirmation gate. The README is explicit that Phase B does not begin until the current proposal is approved, and that a global change (a new aspect ratio or theme) sends the Skill back to the proposal stage rather than mechanically swapping a label. That is a real design decision: it costs you a round trip in the conversation and buys you the ability to fix the story while changes are still cheap.
Aspect ratio is treated as a directing problem, not a formatting one. For `9:16` the README lists vertical depth, a bold central silhouette, layered reveals, and phone-legible text. For `16:9` it lists horizontal staging, lateral camera movement, split-screen comparison, and generous negative space. For `1:1`, strong central composition, radial motion, and clear edge margins. The output is documented as six distinct scenes with a visible change every two to three seconds, an English narration of roughly 130 to 150 words, exact English lines with reference translations, and negative constraints inside each prompt.
Installing the Skill and running a first request
The README gives a three-command install. Clone the repository, then copy the installable Skill folder into your Codex skills directory. The `${CODEX_HOME:-$HOME/.codex}` expansion is the part to read carefully: if `CODEX_HOME` is unset, it falls back to `~/.codex/skills/`.
git clone https://github.com/kaomei/stickman-video-director.git
cd stickman-video-directorcp -R skills/directing-stickman-videos "${CODEX_HOME:-$HOME/.codex}/skills/"After that, restart Codex so the Skill appears in the available list, then invoke it and paste your material. The README's example invocation is a single line:
$directing-stickman-videosThe README also shows a full example request, which combines the invocation with the copy you want turned into a video:
Use $directing-stickman-videos to turn this copy into a one-minute English stickman video:
Gravity bends space and time so strongly around a black hole that even light cannot escape.What you should see next is not a prompt package. The Skill asks for the missing aspect ratio and theme first, then presents the six-act director's proposal for confirmation, and it does not generate the final model prompts before that approval.
Where the Skill stops and you start
Two limitations are stated in the README rather than discovered later. The first is audio. Independently generated videos can show slight differences in voice or music. For maximum consistency the README suggests keeping per-segment synchronized sound effects and using one continuous external voice-over and BGM at stitch time. That is an admission that the six prompts do not fully solve continuity on their own, and it puts a manual editing step back on your desk.
The second is scope. The Skill produces prompts, not video. Nothing in the repository renders or assembles the six clips; the README's fifth step is you stitching them. If you expected a tool that takes copy in and hands a finished file back, this is the wrong shape of tool.
There is a platform constraint too. The Skill is a Codex Skill, so it assumes a Codex environment and a skills directory. Teams working in other editors or in a browser-based generation console get nothing from the install instructions. And because the prompts target Gemini Omni Flash specifically, the output is written for that model's prompt conventions; the README does not claim portability to other video models.
Stickman Video Director compared with prompt templates
The obvious alternative is a saved prompt template or a prompt library pasted into a chat window. The difference is in sequencing. A template gives you one prompt and leaves the story structure to you; Stickman Video Director forces the structure to exist first, as a six-act proposal with an English voice-over and a reference translation, and refuses to emit final prompts until you approve it. A template also tends to be ratio-agnostic, while this Skill redesigns composition, camera path, and text placement per aspect ratio.
The trade-off runs the other way as well. A template has no install step, no Codex dependency, and no skills directory to maintain. If your copy is already well structured, or you only need a single clip, the Skill's confirmation round trip is overhead you are paying for nothing. It earns its place when the material is rough and the minute has to hold together.
Maintenance, licence, and what upgrading costs
The repository is not archived, and the last push was on 2026-08-20. The primary language is Shell, which matches the install instructions: the Skill is a folder you copy, not a package you build. The README points contributors at `tests/` for behavior scenarios and verification scripts, and at `docs/superpowers/specs/` and `docs/superpowers/plans/` for the confirmed product design and implementation plan, so the design rationale is in the tree rather than only in the README.
Upgrade cost is low by construction. There is no dependency manifest to reconcile and no service to redeploy; you re-copy `skills/directing-stickman-videos` and restart Codex. The real cost is re-approving proposals if a Skill update changes how the six-act structure is composed, because the confirmation gate means you will read the proposal again. The licence is MIT, which permits commercial use and modification; the repository ships a `LICENSE` file at the top level. That is a factual note about the licence text, not legal advice, and anyone distributing a modified Skill should read the MIT terms themselves.
Editorial conclusion
Adopt it if you already write in Codex and want the story structure settled before you spend generation credits; the confirmation gate is the whole product. Skip it if you need a rendered video straight out of the tool, if you work outside Codex, or if you need frame-accurate audio continuity across the six clips, since the README admits independently generated segments can differ in voice and music. Before you commit, verify three things: that your Codex build exposes skills, that the copy of skills/directing-stickman-videos lands in your skills directory, and that your target platform matches one of the three supported aspect ratios.
Frequently asked questions
What does Stickman Video Director do?
It is a Codex Skill that turns your copy into a confirmed six-act director's proposal and then into six separate Gemini Omni Flash prompts for a one-minute stickman video. You stitch the six roughly ten-second clips yourself.
Is Stickman Video Director ok for kids?
The README does not address audience suitability or content ratings. It describes the Skill as suited to knowledge explainers, motivational stories, educational shorts, and fast-paced visual content, but makes no claim about child-appropriate output.
What does Stick Man mean in Stickman Video Director?
The README does not define the term as slang. The stickman refers to the visual style of the generated video, shown in two high-contrast variants: a black figure on a white canvas, or a white figure on a black canvas.
What is the best software to make Stickman animation with Stickman Video Director?
The Skill targets Gemini Omni Flash and emits prompts written for it. The README states no API and no MCP dependency are required, so the only software prerequisite it names is a Codex environment with a skills directory.
Community notes