Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add kangarooking/director-skills --skill action-fight-promptgit clone --depth 1 https://github.com/kangarooking/director-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kangarooking/director-skills/action-fight-prompt)<a href="https://agentmods.dev/skills/kangarooking/director-skills/action-fight-prompt"><img src="https://agentmods.dev/badge/skills/kangarooking/director-skills/action-fight-prompt/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kangarooking/director-skills/action-fight-prompt"><img src="https://agentmods.dev/badge/skills/kangarooking/director-skills/action-fight-prompt.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00179 | $0.02229 |
| Opus 5.5 | $0.00072 | $0.00892 |
| Sonnet 5.5 | $0.00036 | $0.00446 |
| Haiku 4.5 | $0.00018 | $0.00223 |
Grade A, and why
action-fight-prompt scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
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:
- 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。 - 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。
- 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。
- 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":
- 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。
- 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。
- 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。
- 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。
- 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。
- 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。
- 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)
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 95 lines · 179 tokens per session scan A d708e75f2840
action-fight-prompt is a skill published in the GitHub repository kangarooking/director-skills (215 stars, last pushed today), licensed MIT. It adds 179 tokens to every session and 2,229 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-19.
Other skills, from other repositories
seedance-prompt
This skill should be used when the user asks to write, improve, translate, compress, or debug a Seedance 2.0 video prompt; mentions T2V, I2V, V2V, R2V, camera direction, prompt quality, or provides reference assets for a production-ready prompt.
lanshu-create-ai-presenter-video
Turn a topic or finished script into a complete, publish-ready explainer video — led by an AI presenter from an authorized adult presenter image, or performed with no presenter in one of twelve explainer styles (nine visual styles and three narrative ones). Not for promos, ads, or footage montages. Use for new…
video-podcast-maker
Use when the user gives a topic and wants an automated topic-driven narrated explainer, podcast, or knowledge-summary video (Bilibili / YouTube / Xiaohongshu / Douyin / WeChat Channels), or asks to learn visual design patterns from a reference video/image. Trigger when the user mentions creating a knowledge video…
video-podcast-maker-lite
Minimal personal narrated-video pipeline — a topic becomes a talking-head-free explainer MP4 (1080p or 4K) via script → Azure TTS (SSML) → Remotion. Use when the user wants a quick narrated video from a topic without the full video-podcast-maker machinery (no extra skills, no thumbnails/shorts/publish matrix). Do NOT…
video-podcast-maker-nano
Smallest personal narrated-explainer-video pipeline (spoken narration over visuals, not an audio podcast), fully tool-agnostic and autonomous by default — topic → research ∥ asset collection → script → TTS → video → 4K render ∥ publish info + cover. The skill defines the pipeline logic and self-verified checkpoints…
prop-asset
A preset for designing reference images of props, meaning objects used in a scene or production. It records an object’s appearance, materials, dimensions, condition, views, ownership, orientation, and consistency rules.