Skip to main content
skill CX Codex

directing-stickman-videos — Claude Code skill: what it does

directing-stickman-videos — a Claude Code skill turning topics, articles, and scripts into animated videos. Director mode builds storyboards, while Autopilot mode produces narrated clips.

MIT 774 installs ★ 7,352 updated 2026-11-16

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add kaomei/stickman-video-director --skill directing-stickman-videos -a claude-code
Open on GitHub

What directing-stickman-videos does

The skill handles video generation through Director and Autopilot modes using generation tools and a local renderer. Director mode builds approved storyboards and standalone prompts across various visual styles like rubber-hose animation. Autopilot mode assembles narrated and captioned clips with minimal line art, retro television aesthetics, or character explainers.

Running the skill requires selecting a workflow path and configuring the generation tools inside the kaomei/stickman-video-director repository.

What it helps with

  • turn text articles into video storyboards and prompts
  • produce narrated animated clips using autopilot mode
  • configure character animation and audio tracks
  • generate meme or retro style animated videos

Example prompts for the agent

  • Turn this article into an animated video using Autopilot mode
  • Create a storyboard and video prompts in rubber-hose style
  • Set up a character explainer scene with narration

How to install directing-stickman-videos

1Skills CLI (Vercel)

npx skills add kaomei/stickman-video-director --skill directing-stickman-videos -a claude-code

2Manually: put the files in place

~/.claude/skills/directing-stickman-videos/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/directing-stickman-videos/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about directing-stickman-videos

How to install directing-stickman-videos in Claude Code?

Run npx skills add kaomei/stickman-video-director --skill directing-stickman-videos -a claude-code, then check ~/.claude/skills/directing-stickman-videos/SKILL.md.

What does directing-stickman-videos do?

It turns topics, requirements, articles, and scripts into stickman videos using Director mode for storyboards and Autopilot mode for automated generation.

What is needed to run directing-stickman-videos?

It requires configured generation tools, a local renderer, and compliance with the workflow reference files inside the repository.

Source text

SKILL.md · MIT
---
name: directing-stickman-videos
description: "Turn topics, requirements, articles or scripts into stickman videos. Director mode creates approved storyboards and standalone video prompts; Autopilot mode produces narrated, captioned illustration-and-micro-animation videos through configured generation tools and a local renderer. Also guide installation handoff, first use and setup of this specific Skill."
---

# Directing Stickman Videos

## Two modes, one Skill

Keep the name and known choices; clarify ambiguity once. Before planning, read [mode execution boundaries](references/mode-execution-boundaries.md), then the selected route:

- **A · Director / A1–A5:** approved proposals and standalone video prompts. A1 maps to Style 1 light/dark, A2 to Style 2/2A/2B, A3–A5 to Styles 3–5. Follow the unchanged contract below.
- **A6 · Amber Noir:** read [A6](references/amber-noir.md) when selected, before planning; bypass the A1–A5 contract below. It owns reference-guided execution, Gemini Puck narration and finishing.
- **A7 · 橡皮管动画风 / Rubber Hose Animation:** read [production](references/rubber-hose.md), [four-block frozen prompts](references/a7-prompt-contract.md), [visual references](assets/a7/reference-manifest.json) and [actual-output acceptance](references/a7-acceptance.md). Default to the successful MV route: fixed character/style blocks + existing kitchen reference material + compact new scene and timed acting/camera → direct video generation. Use the image as an actual identity/drawing reference, not a compulsory kitchen first frame; no mandatory new still or low-resolution probe for every location. Reuse the [two accepted expanded examples](assets/a7/validated-workflow.json), preserving the thin-line two-tuft hero and 1930s ink-and-paint scenery. Review actual motion, contacts, drawing and sound; matching prompts/reference inputs do not certify output. A7 owns its Director workflow and optional narration/captions; sample lyrics are not default requirements. Bypass A1–A5 rules and B renderers.
- **B · Autopilot:** read [Autopilot](references/autopilot.md), then B1 [Minimal Line Art](references/b1-minimal-handdrawn.md), B2 [Colorful Hand-Drawn](references/b2-colorful-handdrawn.md), or B3 [Retro TV](references/b3-television.md). B3 also requires [semantic lines and purposeful motion](references/b3-semantic-motion.md): plan cue-linked events and object-bound connectors automatically, then independently review every event/graphic in the actual MP4; geometry and pixel change do not establish meaning. B1/B2/B3 require [sample-effect acceptance](references/b1-b2-b3-reference-acceptance.md): capable agents target the samples; genuine tool limits permit disclosed best effort, never an equivalence claim.
- **B4 · Survival Thriller:** read [visuals](references/b4-survival-storytelling.md), [asset boundary](references/b4-asset-origin.md), [picture animation](references/b4-picture-animation.md), [audio](references/b4-audio-direction.md), [opening](references/b4-opening-direction.md), [typography](references/b4-typography-acceptance.md) and [QA](references/b4-repeatability-check.md). Create story-specific artwork; references/showcases are quality standards only. Require independent `story_asset_origin` plus all existing reviews. Automatically plan and render story-led local picture changes through the body as well as animated text; no reminder to animate is needed. Preserve the white-glove drawing lock; limb animation is not a completion requirement. Apply the first-three-second visual escalation gate and story-led horror visibility gate, keep every opening/body/ending shot at most three seconds, and time meaningful action within that limit against events/actual speech. Crops, subtitle updates, micro-zooms and decoration do not reset an unchanged principal-picture state’s three-second dwell. Use `b4-hook-v1` for landscape and the [approved portrait poster treatment](references/b4-opening-direction.md#native-portrait-opening) for portrait through the [helper](scripts/b4-opening.mjs). Every language/ratio needs separate independent actual-MP4 picture and text reviews; technical checks are not acceptance.
- **B5 · Character Explainer:** read [B5](references/b5-character-explainer.md) and the [default first-preview workflow](references/b5-first-preview.md). **Before composing a first preview or new story, read the latest [approved filled Mei example](references/b5-mei-filled-example.md), watch its [exact approved film](assets/b5/local-baselines/mei-event-relay/approved-sample.mp4), and read the [actual relay stage and Mei rig](examples/b5-mei-event-relay/README.md). Read the [original Mei prompt](references/b5-mei-original-prompt.md) and fill the [Mei scene template](references/b5-mei-scene-template.md).** Apply [directing structure](references/b5-crazymei-directing-structure.md) and the [motion contract](references/b5-left-stage-motion.md): successive lead visuals and clear handoffs, source proportions, scene richness, and motivated physical action. **CrazyMei + Spider remain the drawing/action standard in both native ratios**; Jensen is historical only. Let the current words choose the ending; never force a side-by-side recap or summary to fill time. [Draw the needed objects, states and contact poses before tweening](references/b5-left-stage-motion.md#draw-needed-states-before-tweening), reusing approved identity/audio and suitable artwork. Use the installed [executable artwork and relay kit](references/b5-art-kit.md), copied by default into each new project: source-drawn contours, complete figure/contact poses, whole cartoon animal states, safe composition and evaluated endpoint handoffs. Prefer graphic substitution and whole-object comic motion over unnecessary animal-joint simulation. Author, render and encode an internal candidate, then inspect that actual candidate once at normal speed with sound, checking motion relay, figures/contact, composition and semantic captions. Inspect suspect frames only as needed; repair and recheck affected intervals and neighbors. Default previews do not require independent storyboard rehearsal, plan-lock, blind PNG/seal/contextual rounds or per-event evidence clips. The [strict formal protocol](references/b5-visual-acceptance.md) remains opt-in. State actual viewing/listening limitations without inventing a pass or replacing unavailable playback with repeated whole-image audits; user acceptance is never inferred.

- **B6 · 动感 Meme 风:** read [B6](references/b6-dynamic-meme.md), the mandatory [installed production workflow](references/b6-installed-production.md) and [character-binding contract](references/b6-character-binding.md). Turn the brief into a finished new film: assets, native shots, binding, sound and QA; a copied gallery or V90 rerender is insufficient. Default init provides compositor, puppet and spatial helpers without exposing the old V90 performance to new-story scenes. Bundled reference/test/example media are compare/study-only, even renamed; paint new story assets. Legacy tests remain diagnostic regardless of fixture/review edits. Explicit unchanged V90 reproduction uses its [recipe](references/b6-hong-kong-v90.md), labelled `example-reproduction`. Native 16:9/9:16 requires separately authored plans; the V90 performance currently validates 16:9 only. For new cast/actions use the installed [executable character packages and compositor](references/b6-puppet-api.md), not numeric pins alone; default projects require their final visible-alpha evidence. Reuse identity/view bindings across scene branches and background cast; verify complete chains, fixed lengths, elbow direction, occlusion, handoffs and grounded legs before extending. For an era/place, apply [period and trend direction](references/b6-art-direction.md#era-place-and-trend-direction) to clothing, hair, accessories and scenery while preserving the brushwork.

B6 authoring defaults to the [faster iteration path](references/b6-fast-iteration.md): check [character imports](references/b6-character-preflight.md) before new paid narration/full encoding, reuse compatible [action timing helpers](references/b6-actions.md), then inspect a true lower-resolution 30fps draft through `scene.draft_frame` and `context.viewport`. Reuse accepted audio; draft/worklist results do not inherit review passes or replace final acceptance.

B6 works best in **portrait (9:16)**. **Sample 1's portrait Eiffel V44 is the strict primary drawing and animation standard**, including its heavier relative limb stroke; Sample 2 V90 is secondary binding/engineering evidence. **Archivo Black is B6’s user-approved default caption font**, retaining the established size, placement, outline and phrase switching. Apply the [Sample 1 acceptance matrix](references/b6-sample-1-acceptance.md) and [motion execution and review gate](references/b6-motion-review.md): lock the primary reference, compare limbs at matched head width, check assets before promising a shot, inspect actual decoded playback with its implementation, and separate technical checks, creative review and user approval. New-story init includes the exact portrait reference, its style standard and the portable spatial helper. A plan label, finer V90 limbs or successful encode does not establish matched quality. All 35 detailed subchecks are mandatory review coverage, including the approved Archivo Black portrait caption profile/placement/raster, semantic timing, head/body paint, all-cast limbs, soft-light/material hierarchy, independent background life, purposeful camera and original-speed sound. An unchanged/synced subtitle is not evidence of matching the approved caption standard. A broad pass cannot hide an unchecked subcheck. Animation performance and spatial richness are mandatory acceptance items. On pending/needs-repair findings, continue authorized local correction, re-rendering and actual review; do not treat a failed test preview as completion or hand the unfinished acceptance work to the user. Preserve and link each rejected version’s repair findings.

A and B share installation, not production contracts. Rules from **A · Director** onward apply only to A1–A5, not to A6, A7 or any B style. Each B owns its visuals, captions, voice, pacing and SFX; no B1 fallback or A setup/clip/approval/text rules. Preserve unrelated styles.

B1–B6: read the [installed workflow](references/b-style-installed-workflow.md) and run its style/ratio query before production.

B accepts topics, briefs, articles and scripts; default English unless overridden. Follow [selected-provider execution](references/autopilot.md#selected-provider-execution): a missing adapter is not a provider prohibition. Use native named TTS without style descriptions; B4 allows secondary-role acting, not narrator tuning. Read the [story-led sound system](references/autopilot-sound-system.md). Never automatically revoice accepted media.

Captions: B1–B4 and B6 single-line; B5 up to three semantic lines. Apply fixed style typography and actual local motion/audio checks in the [common gate](references/autopilot-production-qa.md#natural-captions-and-semantic-sound). B1/B2/B3/B5 voice credits are illustrative; Brian remains B4-only. New B5 real/complex presenters require an available image tool or approved supplied raster, not code-face substitutes; follow the [production gate](references/b5-production-gate.md).

All B styles: native **16:9 and 9:16**, under the [dual-format gate](references/autopilot-production-qa.md#native-landscape-and-portrait). For **B1–B3 portrait**, also read the [native portrait contract](references/b1-b2-b3-native-portrait.md): ratio-specific readable captions, authored per-shot focal regions, safe-area/visible-action checks and independent 1× phone-size review are required. Use the approved internal 10-second B1/B2/B3 portrait samples as effect targets alongside the landscape references; B1 retains its handwritten hierarchy and sequential element builds, while B2/B3 retain their original visual languages. They are not additional README showcases. Their exact-clip user approval, technical diagnostics and each new output's independent review remain separate; an engineering render does not establish portrait acceptance.

## First use and next steps

Read [first-use and stage guidance](references/getting-started.md) after verified installation or on first invocation. Use the conversation/requested guidance language independently of video language or README edition. Reuse choices; do not replay welcome or add approvals. Follow **A5 proactive character guidance** and **Optional A character references** there without changing voice rules.

Read [optional finishing](references/thumbnail-publishing-kit.md) after video approval or a packaging request; offer once, never block delivery.

## A · Director

## Core contract

Turn one source into a confirmed director's proposal and then N standalone prompts for approximately ten-second Gemini Omni Flash clips (where N = target duration in seconds / 10; default 6 clips for 60 seconds). Preserve the source's meaning while structuring it through a high-completion 5-stage emotional narrative arc.

## Setup gate

Require these before planning:

- source material
- aspect ratio: `16:9` or `9:16`
- target duration: in multiples of 10 seconds (e.g. `30s`, `60s`, `90s`, `3min`, `5min`; default: `60s` / 6 clips if unspecified)
- visual style and theme:
  - `Style 1 (Classic Minimalist)`: light (white background, black figure) or dark (black background, white figure)
  - `Style 2 (Modern Beanie Zeke)`:
    - `Style 2A (Modern Studio Tech)`: pure white high-key studio, subtle light-gray perspective grid, floating cyan/blue glass UI
    - `Style 2B (Cinematic Story)`: full-color narrative environments, cinematic lighting & depth
  - `Style 3 (Prehistoric Hand-Drawn / Dark Comedy)`: expressive white filled heads, black stick limbs, animal-hide costumes and earthy prehistoric settings; story-specific casting
  - `Style 4 (Cozy Storytelling / 治愈故事风)`: expressive minimalist white-filled characters with dark stick limbs in full-color 2D cartoon environments; varied story locations and cinematic shot changes
  - `Style 5 (Historical Cartoon / 历史卡通风)`: white heads with connected short white necks, expressive faces, black stick limbs and compact fingerless palms, period dress and cel-shaded historical settings

If aspect ratio or visual style is missing, ask for all missing items in one concise message and stop. If duration is omitted, default to 60 seconds (6 clips). If a user provides a duration not divisible by 10, round to the nearest 10-second multiple. If a user specifies only light or dark theme, default to Style 1. Never select an aspect ratio or style silently. Do not re-ask choices already supplied.

Urgency, generation cost, client pressure, and requests to "pick normal settings" do not waive this gate.

## Workflow

1. Read `references/storyboard-template.md`, `references/style-catalog.md` and `references/voice-continuity.md`. Structure the narrative using the **5-Stage High-Completion Heartbeat Engine** (Golden Hook → Disrupting Assumptions → Unveiling Insider Secrets → Ultimate Truth Revelation → Elevation & Interactive Discussion), with timing dynamically scaled to the target duration. Produce Phase A in the user's language, with English VO and a reference translation.
2. Stop after the director's proposal and request explicit approval.
3. If the user changes ratio, duration, style, theme, narration, scene structure, or global direction, recompose Phase A and request approval again.
4. Only after approval of the current Phase A, read `references/omni-flash-prompt-contract.md` and produce Phase B.
5. Use `references/examples.md` only when a concrete end-to-end example would resolve ambiguity.

Topic approval, schedule pressure, or approval of an older draft is not approval of the current Phase A.

## Output rules

For Style 3, read `references/prehistoric-dark-comedy.md` before planning. Its casting, story rhythm, palette and continuity rules override the generic no-idle, device-count and five-stage discussion-ending rules below. Use setup → escalation → ironic payoff for narrative shorts; do not force a discussion question into a visual punchline. This is a visual storytelling style, not a claim about actual prehistory.

For Style 4, read `references/cozy-storytelling.md` before planning and prompt assembly. Default each new ten-second segment to a different recognizable story setting, unless the user explicitly requests one location or approves a deliberate return. Preserve characters, drawing grammar, important props, movement direction and narrator—not an identical table or background across the whole video. Name the location cuts and any time jumps in Phase A. This is full-color cartoon storytelling, not a hand-drawn/sketchbook mode and not a claim of photorealistic 3D backgrounds.

For Style 5, read `references/historical-cartoon.md` before planning and prompt assembly. Keep its drawing grammar fixed while changing the era pack. Confirm story-specific cast cards, setting, palette, props and exact narration in Phase A; repeat the full approved style and visible cast descriptions in every Phase B prompt. Its historical palette, music, physical-action pacing and story ending override generic three-color, mandatory-piano, device-count, no-idle and discussion-ending rules below. Apply the existing shared voice contract; do not apply this style's character rules to Styles 1–4.

- Structure every script using the 5-stage high-completion flow:
  1. **Golden Hook**: Instant curiosity trigger via a counter-intuitive question or visual paradox. Timing is flexible based on duration (e.g. within the first 2–5 seconds in short videos, up to 10–15s in long videos); never open with generic platitudes.
  2. **Disrupting Assumptions**: State what everyone thinks is true, then shatter it in one sentence to build cognitive conflict.
  3. **Unveiling Insider Secrets**: Pull back the curtain on hidden mechanics, counter-intuitive details, or the real obstacle.
  4. **Ultimate Truth Revelation**: Deliver the underlying business logic or human psychology truth with maximum cognitive payoff.
  5. **Elevation & High-Engagement Discussion**: Conclude with a memorable punchline and an irresistible discussion prompt that sparks debate in the comments.
- Scale English VO word count to approximately 20–25 words per 10-second clip across N clips (e.g. ~60–75 words for 30s / 3 clips, ~120–150 words for 60s / 6 clips, ~360–450 words for 3min / 18 clips, ~600–750 words for 5min / 30 clips).
- Give each clip three timed beats, at least four relevant visual devices, and a visual change every two to three seconds.
- Keep character proportions, line weight, style, and narrator consistent across all clips.
- For Style 1, freeze the concise character anchor from `references/style-catalog.md` and repeat it verbatim in every clip. Keep two arms and two legs, one current pose per limb, stable proportions and uniform medium line weight; a moving arm replaces its previous pose rather than adding another arm. Keep the figure monochrome and saturated accents on props/graphics. Do not replace the anchor with "same character", ellipses, numeric anatomy recipes, skeleton/path instructions or clip-specific appearance descriptions. Occasional simple line expressions are allowed; the head remains hollow. Keep rich actions and effects separate from character design. Treat generated consistency as a visual acceptance check, not a guarantee from repeated text.
- For Style 1, limit the video to three saturated accent colors named with ordinary descriptive words (e.g. vivid red, electric blue, warm gold). For light theme, enforce a flat, digitally pure-white canvas with no textures, gradients, or 3D depth.
- For Style 2 (Modern Beanie Zeke), enforce character consistency:
  - Clip 1: `A minimalist 2D animated stick figure wearing a bright red beanie (smooth knit, no pom-pom) and a yellow t-shirt, with simple black stick limbs and shorts. Simple black lines, vibrant colors, smooth 2D animation style.`
  - Clips 2–N: `The same minimalist 2D animated stick figure in a bright red beanie and yellow shirt... Simple black lines, vibrant colors, smooth 2D animation style.`
  - No detailed eyes, pupils, or photorealistic features (avoids bug-eye deformation).
- For Style 2A (Modern Studio Tech), enforce: `in a modern bright white studio space with subtle light-gray perspective grid lines on the floor plane. High-key studio lighting, clean white negative space, sleek glowing cyan and electric blue glass holographic UI elements.` Always include negative constraint: `STRICTLY MINIMALIST, NO CIRCUIT BOARD TEXTURES, NO SCI-FI WALL PANELS, NO CRACKED CONCRETE.`
- For Style 2B (Cinematic Story), specify full-color narrative setting and cinematic lighting while preserving the 2D animated stick-figure character design.
- For Style 4, freeze each approved character's complete anchor and repeat it in every clip where that role appears. Separate static design from first-frame positions, timed prop transfers and final states; changing scene does not reset prop ownership or the narrator. Read its reference for shot variety and visual acceptance checks.
- For all styles, avoid abstract liquid/shape morphing. Drive motion with concrete character actions (leaping, touching glass, drawing luminous lines, opening doors) with timed beats across `[0–3s]`, `[3–7s]`, and `[7–10s]`. Never leave the character idle.
- For **all styles**, apply `references/voice-continuity.md`: choose a complete video-level voice profile in Phase A, then repeat its full frozen English block verbatim in every clip. Different videos may choose different voices; a scene/style/character change must not recast the narrator. Lock timbre, register, accent, texture, delivery, pace and recording treatment, not just "same narrator". Keep emotional progression in visuals/music. Style-specific music rules never override this shared voice contract. Mandate BGM continuity from clip 1.
- Treat narration as audio-only (`Audio voiceover only, strictly no speech bubbles, no dialogue boxes`). Quote exact dialogue and forbid alteration, repetition, captions, subtitles, or visual transcription.
- Never place technical color notation (hex, RGB, Pantone) inside generation prompts.
- Default generated clips to no visible words, letters, numbers, or interface copy. Put optional two-to-five-word overlays in a separate post-production note.
- Match every clip ending to the next clip opening through action, an important prop, screen direction or a planned cut. Style 4 uses deliberate scene changes or approved time jumps; it does not require the same background in adjacent clips.
- Do not invent unsupported facts, statistics, quotations, or product claims.

## Revision rules

Recompose rather than rename:

- `16:9`: stage action across left, center, and right; use lateral tracking and negative space.
- `9:16`: use depth, stacked motion, vertical reveals, and interface-safe placement.

Theme, ratio, duration, or style changes require a fresh visual balance check. A global change invalidates prior approval.

## Final check

Apply the checklist in the loaded reference. Repair any failed condition before responding.

© kaomei · MIT · kaomei/stickman-video-director

Similar skills

All skills for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open