Agent skill

Action Fight Prompt

by kangarooking in kangarooking/director-skills

动作打戏 AI 视频提示词设计技能。当用户需要设计具备电影级运镜、节奏感、打击感的武打/战斗片段提示词时使用——覆盖双角色对打、女主 vs 怪兽、近身格斗、载具追逐战、群战、一镜到底、巨型生物/机甲大场面等场景。核心方法论:时序拆段(15 秒拆成 2-3 秒阶段)、动作因果链、环境反馈、视觉资产锁定与参考图映射、音效层。触发词:动作打戏、打戏提示词、武打镜头、战斗片段、打击感、fight…

MITAuto-check passedMedia & Creative

Install Action Fight Prompt

skills CLI
$ npx skills add kangarooking/director-skills --skill action-fight-prompt -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install kangarooking/director-skills action-fight-prompt --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/kangarooking/director-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/action-fight-prompt .claude/skills/action-fight-prompt && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
action-fight-prompt
GitHub stars
181
Token cost
~733 tokens
SKILL.md length
130 words
Files
4 (incl. references)
Skills in repo
6
Repo updated
First seen
Licence
MIT

At a glance

动作打戏 AI 视频提示词设计技能。当用户需要设计具备电影级运镜、节奏感、打击感的武打/战斗片段提示词时使用——覆盖双角色对打、女主 vs 怪兽、近身格斗、载具追逐战、群战、一镜到底、巨型生物/机甲大场面等场景。核心方法论:时序拆段(15 秒拆成 2-3 秒阶段)、动作因果链、环境反馈、视觉资产锁定与参考图映射、音效层。触发词:动作打戏、打戏提示词、武打镜头、战斗片段、打击感、fight…

  • Works in 4 steps: 时序拆段:总时长(≤15 秒)拆成多个 2-3… → 动作因果链:相邻阶段的动作必须是物理因果关系——对手前冲→女主借惯性转身;女主击中… → 环境反馈:每个阶段至少写一处环境反馈(尘土飞扬、金属撞击、沙石崩塌、地面裂纹、碎屑… → …
  • Media & Creative work in your project
  • SKILL.md covers 四条铁律(缺一不可), 进阶规则(决定成片质感上限), 输出样式(二选一,按题材) and 工作流程, plus 1 more section
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Action Fight Prompt is an agent skill from kangarooking/director-skills. 动作打戏 AI 视频提示词设计技能。当用户需要设计具备电影级运镜、节奏感、打击感的武打/战斗片段提示词时使用——覆盖双角色对打、女主 vs 怪兽、近身格斗、载具追逐战、群战、一镜到底、巨型生物/机甲大场面等场景。核心方法论:时序拆段(15 秒拆成 2-3 秒阶段)、动作因果链、环境反馈、视觉资产锁定与参考图映射、音效层。触发词:动作打戏、打戏提示词、武打镜头、战斗片段、打击感、fight scene、对打、动作设计、追车战、群殴、一镜到底。

Its SKILL.md is about 730 tokens, which your agent loads only when the skill is triggered. The skill folder holds 4 other files, including reference files (for example `references/camera-and-impact.md`, `references/example-15s-duel.md` and `references/scene-patterns.md`).

It sits in Media & Creative. The repository describes itself as: 导演Skill:面向 AI 视频创作的开源 Agent Skills | Director Skills: Open-source Agent Skills for AI video creation. The licence is MIT.

When your agent uses it

  • Media & Creative work in your project

Example prompts

  • “/action-fight-prompt”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. 时序拆段:总时长(≤15 秒)拆成多个 2-3 秒的小阶段。每个阶段只负责一组主要动作,用时间戳显式标出(如 0:00-0:02)。禁止一个提示词塞十几个同时发生的动作。
  2. 动作因果链:相邻阶段的动作必须是物理因果关系——对手前冲→女主借惯性转身;女主击中腹部→对手身体前倾→下一段才可能抓住肩膀。每个动作都要能回答"上一个动作给了它什么条件"。孤立动作 = 瞬移/穿模的根源。
  3. 环境反馈:每个阶段至少写一处环境反馈(尘土飞扬、金属撞击、沙石崩塌、地面裂纹、碎屑飞溅)。这不是装饰——它告诉模型力量来自哪里,打击感由此落地。
  4. 资产锁定:提示词开头用独立段落声明人物的脸、发型、服装、武器、身体比例全程不变。双角色场景必须分别说明每张参考图对应谁(图1 = 女主,图2 = 怪兽),防止结构混淆。

What it can do on your machine

Read from SKILL.md and the folder at commit 459deba. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Action Fight Prompt loads about 733 tokens when it runs, and up to ~4k if it reads all its reference files. Until then it costs about 61 tokens; SKILL.md has 130 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~61
When it runs · the whole SKILL.md, loaded when a task matches
~733
With references · SKILL.md plus every file in references/, read only if the agent opens them
~4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from kangarooking/director-skills at commit 459deba, republished under its MIT licence (© kangarooking). 130 words, ~733 tokens.

Download SKILL.mdSave it as .claude/skills/action-fight-prompt/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
name
action-fight-prompt
description
动作打戏 AI 视频提示词设计技能。当用户需要设计具备电影级运镜、节奏感、打击感的武打/战斗片段提示词时使用——覆盖双角色对打、女主 vs 怪兽、近身格斗、载具追逐战、群战、一镜到底、巨型生物/机甲大场面等场景。核心方法论:时序拆段(15 秒拆成 2-3 秒阶段)、动作因果链、环境反馈、视觉资产锁定与参考图映射、音效层。触发词:动作打戏、打戏提示词、武打镜头、战斗片段、打击感、fight scene、对打、动作设计、追车战、群殴、一镜到底。

动作打戏提示词设计

为 AI 视频生成(miora 等通道)设计动作打戏提示词。目标:电影级运镜、清晰的节奏感、可信的打击感,杜绝瞬移、穿模、无效挥拳。

四条铁律(缺一不可)

设计任何打戏片段前,先按这四条搭骨架,再填血肉:

  1. 时序拆段:总时长(≤15 秒)拆成多个 2-3 秒的小阶段。每个阶段只负责一组主要动作,用时间戳显式标出(如 0:00-0:02)。禁止一个提示词塞十几个同时发生的动作。
  2. 动作因果链:相邻阶段的动作必须是物理因果关系——对手前冲→女主借惯性转身;女主击中腹部→对手身体前倾→下一段才可能抓住肩膀。每个动作都要能回答"上一个动作给了它什么条件"。孤立动作 = 瞬移/穿模的根源。
  3. 环境反馈:每个阶段至少写一处环境反馈(尘土飞扬、金属撞击、沙石崩塌、地面裂纹、碎屑飞溅)。这不是装饰——它告诉模型力量来自哪里,打击感由此落地。
  4. 资产锁定:提示词开头用独立段落声明人物的脸、发型、服装、武器、身体比例全程不变。双角色场景必须分别说明每张参考图对应谁(图1 = 女主,图2 = 怪兽),防止结构混淆。

进阶规则(决定成片质感上限)

四条铁律保证"能看",以下规则保证"好看":

  1. 音效独立成层:单独写音效段——要么每阶段末尾跟一小句音效,要么结尾整段"音效节奏"总述。无配乐需求时显式写"不要音乐,只保留音效/环境音"。音效要覆盖力量(引擎拔高、闷响)、材质(碎石、摩擦)、生物(嘶吼、呼吸)三层。
  2. 动作结果允许失败:打击 ≠ 命中即胜。刀砍不进硬甲被弹开、反作用力震飞主角、刀刃卷曲——物理一致的挫败比全程碾压更有可信度,也为战术转变提供叙事钩子。
  3. 动作轨迹拆解式描写:关键一击不写结果、写轨迹——"空中下落 → 双手高举 → 竖向重劈 → 刀锋正中厚甲",四步链条让模型逐帧可执行。
  4. 微慢动作给具体时长:命中瞬间的慢速不写"轻微慢动作",写"约 0.2-0.3 秒微慢动作后立刻恢复正常速度",每段最多一次。
  5. 结尾必须铺垫下一轮:余韵段不止定帧——车头重新朝向目标、意识到破不了甲、扯下工牌走向出口。给下一段留钩子。
  6. 群体敌人分层写:多敌人场景交代数量与行为顺序("6 只伴猎怪纷纷转身嘶吼,其中 2 只率先扑出"),镜头明确拍出该顺序;群战中多对手面孔必须各不相同、禁止与主角相似。
  7. 风格开关前置:一镜到底、手持手机纪实质感、无防抖抖动等整体风格声明放在提示词最前部,只声明一次,不做中途切换。

输出样式(二选一,按题材)

  • A. 时间戳式:0:00-0:02 段落正文。适合快节奏近身对打、追车战——节奏密集,镜头切换频繁。正文内镜头语言以"镜头用……跟拍"句式自然嵌入。
  • B. 分镜头命名式:镜头01——接近。适合大场面、景别大幅跳转(贴海浪跟踪 → 环绕 → 极远航拍),每个镜头一个动词性命名,段落内写透单一景别。

两种样式都遵守四条铁律,同一条提示词内不混用。

工作流程

第一步:收集输入(缺项一次性问全)
  • 总时长(miora H3 通道为 4-15 秒整数;区间外不提交,向用户说明)
  • 角色:几方?人类/怪兽?各自的武器?有无载具?
  • 参考图:有几张,各对应谁
  • 场景环境:地面材质(沙石/金属/废墟/水面)——决定环境反馈词汇
  • 题材类型:近身对打 / 载具追逐 / 群战 / 巨型生物对决——决定用哪套场景配方(见 references/scene-patterns.md)与输出样式
  • 音效要求:是否需要配乐(默认不要音乐、只保留音效)
第二步:搭节奏骨架

按"三拍式"分配阶段,再微调:

拍占比内容
蓄力~30%对峙、逼近、第一次交锋试探
爆发~50%主交锋:连续因果动作链,打击最密集
余韵~20%收尾定帧:胜负分明、尘埃落定、镜头缓推

输出阶段表:每阶段填 4 列——时间戳 / 镜头 / 主动作(一组)/ 环境反馈。先给用户看这张表,确认后再展开成完整提示词。

第三步:写完整提示词

按此顺序组装(可选段按需取用):

【风格开关段】一镜到底/手持纪实质感等整体声明(如有)
【资产锁定段】人物外观声明 + 参考图映射;群战加对手面孔规则
【场景段】环境、光线、色调、氛围粒子(尘埃/水雾/粉笔灰)
【时序段】逐阶段:时间戳或镜头命名 + 镜头语言 + 动作 + 因果衔接词 + 环境反馈
【音效段】分段音效或整体"音效节奏"总述;默认声明"不要音乐"
【技术段】画幅、帧率、画质参数(16:9、24 FPS 等按项目规范)

四条铁律的词库与范例见 references/camera-and-impact.md;四类题材配方与两种输出样式骨架见 references/scene-patterns.md;完整范例(女主 vs 怪兽 15 秒)见 references/example-15s-duel.md。

第四步:QA 核对

交付前逐项自查,不通过就改:

  • 每阶段 ≤ 一组主动作,时长 2-3 秒
  • 任意相邻两阶段能说出因果衔接("因为 A 所以 B"成立)
  • 每阶段 ≥ 1 处环境反馈
  • 开头有资产锁定段;双角色场景参考图映射明确;群战对手面孔规则齐全
  • 运镜有变化(不是全程固定机位),且镜头服务于动作(爆发段贴身跟拍/手持震动,余韵段缓推或定帧)
  • 无孤立动作、无"瞬移式"位移描写
  • 关键一击写了动作轨迹拆解;命中瞬间慢速给了具体时长(≤一次)
  • 有独立音效层,且"不要音乐"约束按需求写明
  • 结尾有下一轮铺垫(转向目标 / 战术转变 / 情绪定格)
  • 输出样式全文统一(时间戳式或分镜头命名式,不混用)

边界

  • 只设计提示词文本,不提交生成任务(生成走 miora-video-studio 等对应技能)。
  • 时长超出通道上限时不拆分提交——是否分镜分批由用户决定。

© kangarooking, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in action-fight-prompt of kangarooking/director-skills.

  • SKILL.md
  • references/camera-and-impact.md
  • references/example-15s-duel.md
  • references/scene-patterns.md

Open the folder on GitHubat commit 459deba

Compare with similar skills

Action Fight Prompt next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Action Fight Prompt compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Action Fight Prompt this skillkangarooking/director-skills181—~733Automated safety check: PassMIT
Guizang Social Cardsop7418/guizang-social-card-skill7.4k1 repos~7.8kAutomated safety check: PassAGPL-3.0
Weekly Changelog Videoheygen-com/hyperframes60k—~3.3kAutomated safety check: PassApache-2.0
Anthropic Brand Stylinganthropics/skills180k30 repos~559Automated safety check: PassApache-2.0
MoneyPrinterTurbo Video Generatorharry0703/MoneyPrinterTurbo130k—~2.1kAutomated safety check: WarnMIT
HyperFrames Media Useheygen-com/hyperframes60k—~2.4kAutomated safety check: PassApache-2.0

Similar skills

  • Guizang Social Cards

    op7418/guizang-social-card-skill

    Produces social card sets for Xiaohongshu and WeChat: carousels, Live Photo motion cards and puzzle layouts, and WeChat cover pairs, rendered from single-file HTML.

    7.4k GitHub starsUsed in 1 repo~7.8k tokens
    Media & CreativeAuto-check passed
  • Weekly Changelog Video

    heygen-com/hyperframes

    Turns a weekly changelog markdown file into a branded HyperFrames video with voiceover, animated mock-UI scenes and captions, using fonts, background and scripts bundled in the skill.

    60k GitHub stars~3.3k tokensUpdated today
    Media & CreativeAuto-check passed
  • Anthropic Brand Styling

    anthropics/skills

    Official

    Applies Anthropic's brand colors and fonts to artifacts such as PowerPoint slides, using fixed hex values for text and accents, Poppins headings and Lora body text.

    180k GitHub starsUsed in 30 repos~559 tokens
    Media & CreativeAuto-check passed
  • MoneyPrinterTurbo Video Generator

    harry0703/MoneyPrinterTurbo

    Installs and runs MoneyPrinterTurbo to turn a topic or script into a finished short video with voice-over, subtitles, stock footage and music.

    130k GitHub stars~2.1k tokensUpdated yesterday
    Media & CreativeAuto-check: warnings
  • HyperFrames Media Use

    heygen-com/hyperframes

    Finds, generates and edits media for HyperFrames video projects: music, sound effects, images, icons, logos, voiceovers, captions and color grades.

    60k GitHub stars~2.4k tokensUpdated today
    Media & CreativeAuto-check passed
  • Holo Card Studio

    EverettFish/holo-card-studio

    Create collectible holographic foil cards and two-image lenticular flip cards with AI-generated full-color ukiyo-e and colored sumi-e anime artwork, layered Blender scenes, renders, GLB export, and…

    1.9k GitHub stars~1.4k tokensUpdated 19 days ago
    Media & CreativeAuto-check passed

More from kangarooking/director-skills

  • Travel Skill

    kangarooking/director-skills

    规划、编写、审阅和修复由真实授权素材与 AI 生成镜头混合制作的电影级文旅宣传片。用于文旅短片创意简报、场景图筛选、首帧空间分析、分镜脚本、画面内容提示词、人物动作、导演级运镜、自然动机布光、转场与尾帧衔接、Seedance 2.0/可灵/Midjourney 提示词、生成结果诊断、时间线与合规风险检查;尤其适用于需要写实、辽阔、自由感和角色剧情连续性的项目。

    181 GitHub stars~756 tokensUpdated 4 days ago
    Auto-check passed
  • Miora Video Studio

    kangarooking/director-skills

    miora 视频生成的可靠性与规格层——补内置技能未覆盖的部分:四必填参数闸门(模式/分辨率/时长/画幅缺一不可提交)、时长区间闸门、完成信号的判定与轮询等待、官方规格口径与本通道实测差异(时长/分辨率/参考图上限)、文生/首尾帧/参考生三种模式的参数事实、提示词的自动修正边界、生成状态向用户的同步。当用户要生成视频、动效、短视频,或提到参考生视频/首尾帧/图生视频、miora…

    181 GitHub stars~4.6k tokensUpdated 4 days ago
    Auto-check passed
  • Emotion Performance Director

    kangarooking/director-skills

    A skill your agent uses when the user wants to design, rewrite, refine, or vary emotional acting prompts for live-action AI video characters.

    181 GitHub stars~2.4k tokensUpdated 4 days ago
    Auto-check passed
  • Cinematic Asset Prompts

    kangarooking/director-skills

    影视概念美术专家 & 视觉导演技能——从剧本/故事/世界观文字中提取并设计 AI 视频所需的标准化视觉资产提示词。当用户要求"从剧本提取角色/场景/道具/载具提示词"、"生成角色定妆照 prompt"、"生成场景空镜 prompt"、"生成道具/载具/机械资产设定图 prompt"、"为 AI…

    181 GitHub stars~1.1k tokensUpdated 4 days ago
    Auto-check passed
  • Travel Photo To Video

    kangarooking/director-skills

    根据用户提供的 1–10 张真实旅拍照片和旅游地点,设计文旅宣传片的推荐分镜顺序、统一视觉母版与逐镜头首帧图生视频提示词。适用于让旅拍照片自然动起来、设计克制运镜和后期剪辑连续性的请求;输出可独立复制的提示词,不直接提交视频生成任务。

    181 GitHub stars~1.6k tokensUpdated 4 days ago
    Auto-check passed

Questions about Action Fight Prompt

What does Action Fight Prompt do?

动作打戏 AI 视频提示词设计技能。当用户需要设计具备电影级运镜、节奏感、打击感的武打/战斗片段提示词时使用——覆盖双角色对打、女主 vs 怪兽、近身格斗、载具追逐战、群战、一镜到底、巨型生物/机甲大场面等场景。核心方法论:时序拆段(15 秒拆成 2-3 秒阶段)、动作因果链、环境反馈、视觉资产锁定与参考图映射、音效层。触发词:动作打戏、打戏提示词、武打镜头、战斗片段、打击感、fight…. Action Fight Prompt is an agent skill from kangarooking/director-skills.

When should I use Action Fight Prompt?

Action Fight Prompt fits situations like: media & Creative work in your project.

How do I install Action Fight Prompt in Claude Code?

Run `npx skills add kangarooking/director-skills --skill action-fight-prompt -a claude-code`. Or copy the skill folder (action-fight-prompt in kangarooking/director-skills) into .claude/skills/action-fight-prompt in your project. Claude Code loads it when a task matches its description.

How do I install Action Fight Prompt in Codex?

Run `npx skills add kangarooking/director-skills --skill action-fight-prompt -a codex`. Or copy the skill folder (action-fight-prompt in kangarooking/director-skills) into .agents/skills/action-fight-prompt in your project. Codex loads it when a task matches its description.

Can I use Action Fight Prompt in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add kangarooking/director-skills --skill action-fight-prompt -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/action-fight-prompt, .gemini/skills/action-fight-prompt, .github/skills/action-fight-prompt and .opencode/skills/action-fight-prompt in your project.

What does Action Fight Prompt need to run?

SKILL.md names no scripts, command-line tools or credentials: Action Fight Prompt is instructions for the agent only.

Does Action Fight Prompt access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Action Fight Prompt safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Action Fight Prompt use?

Action Fight Prompt is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Action Fight Prompt use?

About 733 tokens (SKILL.md is roughly 2.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 3.2k tokens, read only when the agent opens those files.

What are the alternatives to Action Fight Prompt?

Skills that share tags, products or a category with Action Fight Prompt: Guizang Social Cards (op7418/guizang-social-card-skill, 7.4k stars), Weekly Changelog Video (heygen-com/hyperframes, 60k stars), Anthropic Brand Styling (anthropics/skills, 180k stars) and MoneyPrinterTurbo Video Generator (harry0703/MoneyPrinterTurbo, 130k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Action Fight Prompt?

kangarooking (a GitHub user) maintains it in kangarooking/director-skills, which has 181 GitHub stars. The repository holds 6 skills in this directory. The repository was last updated on October 7, 2026.

Source: kangarooking/director-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.