Skip to content

doublesq97-ui/su-ai-short-drama / skills/ai-short-drama-studio-male

ai-short-drama-studio-male

Verified

Develop and direct male-audience AI short dramas through role-specific workflows for showrunning, screenwriting, casting and original character design, art, performance, cinematography, storyboarding, sound, post, continuity, and asset handoff. Use when the user asks for 男频短剧、微短剧、玄幻修仙、都市逆袭、职业创业、历史权谋、末世科幻、悬疑规则、竞技成长、选题、剧本、改稿、选角定妆、分镜运镜、生图/生视频制作说明、项目圣经或跨工种协作。

references/showrunning.md

View on GitHub
# Showrunning and total direction

Use this reference for end-to-end planning, cross-craft decisions, production gates, and continuing an existing project.

## Choose the operating mode

### New project

1. Read all supplied story material.
2. Confirm `male` or `neutral/mixed` and load the male story engine only when its viewing promises apply. If the project is clearly female-audience, route it to the female edition instead of improvising an absent branch.
3. Produce a compact development diagnosis: story promise, protagonist goal, opposition, escalation path, differentiator, production risks, and missing decisions.
4. Confirm only high-impact choices, then establish the project bible.
5. Move through script, character, scene, storyboard, generation, and post gates in order.

Completion criterion: the project has a clear viewing promise, a viable production route, named source-of-truth artifacts, and one next work package.

Use [development-review.md](../assets/templates/development-review.md) when the user asks for a formal development audit, gate verdict, or cross-craft work order.

### Continue an existing project

1. Read the project bible and its latest change record.
2. Locate the active work package, locked input, current asset versions, and open blockers.
3. Resume at that gate; do not repeat approved development or regenerate locked assets.
4. Reconcile any conversation instruction that conflicts with the bible before execution.

Completion criterion: continuation starts from the latest confirmed state without duplicating or overwriting completed work.

### Single-craft request

Route directly to the named role. Supply the smallest context needed and return a handoff only if the user is running a sustained project.

## Production spine

### Gate 1: development lock

Lock or explicitly leave open:

- audience branch, genre combination, format, platform, length, and rating constraints;
- logline, protagonist objective, principal opposition, story world, viewing promise, and ending direction;
- production differentiator and likely cost/complexity drivers.

Reject a premise that only names tropes without a specific protagonist decision and consequence chain.

### Gate 2: story lock

Require:

- a whole-story phase map;
- character arcs and relationship changes;
- episode objectives, state changes, setups, payoffs, and hook rotation;
- a locked episode script for the current production batch.

Use a batch size the user can review responsibly. Treat one-to-five episodes as a practical default, not a universal rule.

### Gate 3: preproduction lock

Build a dependency-ordered prep list:

1. characters required by the current batch;
2. recurring and story-critical scenes;
3. interactive props, wardrobe states, injuries, makeup changes, vehicles, creatures, interfaces, and written information;
4. sound, voice, caption, music, and post needs;
5. shots whose complexity may require rewriting, splitting, practical alternatives, or extra reference assets.

Separate **must lock before production** from **can resolve during shot design**. Do not create an exhaustive asset library for material not yet scheduled.

### Gate 4: asset lock

For every character and recurring scene, require:

- one selected identity/version;
- a continuity card with observable anchors;
- stage-specific variants;
- file path, code, source/authorization status, and lock date;
- explicit approval from the project's authorized reviewer, or authorization for a provisional test.

Candidate images remain candidates until this gate passes.

### Gate 5: storyboard lock

Check the current episode as a whole before approving individual shots:

- visual and emotional rhythm varies with story function;
- shot count follows content and model/production feasibility rather than a fixed quota;
- action chains, dialogue, reaction, props, entry/exit states, and transitions connect;
- every tool-specific instruction is derived from a tool-neutral shot decision;
- generation complexity is concentrated where it adds audience value.

### Gate 6: generation and take selection

Treat generation as production, not approval:

- generate bounded candidates;
- compare identity, spatial, action, temporal, sound, and style continuity;
- log defects and select a take or revise the shot;
- never promote an output merely because generation succeeded.

### Gate 7: post and release

Confirm edit rhythm, dialogue intelligibility, OS placement, sound layers, captions, visual text, music rights, asset rights/source records, and continuity across the released cut. Update the project bible to the actual deliverable.

## Showrunner decision rules

- Protect the viewing promise before protecting a favorite trope or shot.
- Solve a recurring error at its source: script, asset, shot design, tool adapter, or post—not with repeated negative prompt clutter.
- Prefer a small locked production batch over a large unreviewed inventory.
- Expose meaningful creative alternatives before lock; reduce choice after lock.
- Preserve role boundaries. Request a rewrite when the story is broken, a new asset when identity is broken, and a new take when execution is broken.
- Keep common craft rules tool-neutral. Place current model syntax and limitations only in `tool-adapters.md` or the project bible with a verification date.

## Production-risk review

Before committing a batch, mark each item low, medium, or high risk:

- number of speaking characters in one continuous action;
- complex contact, fighting, eating, object transfer, vehicles, animals, crowds, or transformations;
- readable screens, documents, signage, subtitles, or exact logos;
- character identity changes, age changes, injuries, wetness, dirt, or wardrobe transitions;
- long synchronized dialogue, overlapping speech, or precise lip sync;
- spatial continuity across multiple entrances/exits;
- period, professional, legal, medical, military, or cultural accuracy;
- references whose authorization or consent is not established.

For each high-risk item, choose one response: improve references, simplify staging, split the action, move the task to post, verify specialist facts, or revise the script while preserving intent.

## Episode-level approval

Approve the episode only when:

- the opening creates an immediate question, pressure, or promise appropriate to the tone;
- each scene changes information, power, relationship, resource, risk, or commitment;
- the protagonist makes at least one consequential choice;
- the ending pays something off while creating a non-repetitive next question;
- the requested audience engine is visible in causality and payoff;
- production requirements fit the available time, tools, assets, and quality bar;
- the next role receives a complete, versioned input.
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 323kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 198kUpdated Claude CodeCodexVerifiedInstall