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 62656456/ai-film-skills --skill whitebox-previs-executorgit clone --depth 1 https://github.com/62656456/ai-film-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/62656456/ai-film-skills/whitebox-previs-executor)<a href="https://agentmods.dev/skills/62656456/ai-film-skills/whitebox-previs-executor"><img src="https://agentmods.dev/badge/skills/62656456/ai-film-skills/whitebox-previs-executor/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/62656456/ai-film-skills/whitebox-previs-executor"><img src="https://agentmods.dev/badge/skills/62656456/ai-film-skills/whitebox-previs-executor.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 | $0.00117 | $0.01993 |
| Opus 5 | $0.00059 | $0.00996 |
| Sonnet 5 | $0.00023 | $0.00399 |
| Haiku 4.5 | $0.00012 | $0.00199 |
Grade A, and why
whitebox-previs-executor 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 4d 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 — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
提示词 3D 白模预演器
Public experimental package. This package is an explicit test executor, not a stable promise of arbitrary choreography. It contains the twelve runtime files only; the local workbench and Blender binaries are not bundled.
可配置运行依赖
- 普通或关节预演需要已安装的 Blender。定位顺序是
--blender、WHITEBOX_BLENDER环境变量、PATH，最后是当前任务下的_runtime/blender-*-windows-x64/blender.exe。指定的路径无效时直接报错，不暗换版本。 - 两个启动器要求显式
--runtime-dir，并将子进程临时目录及应用配置定向其中；在结构、后端路由或依赖检查失败时不创建运行缓存。 - 历史软件渲染器的 FFmpeg 定位顺序为
--ffmpeg、WHITEBOX_FFMPEG、PATH；探测器为--ffprobe、WHITEBOX_FFPROBE、PATH或 FFmpeg 同目录的对应可执行文件。公开包不查找作者电脑的软件路径。 - 相较本地副本，公开包仅增加依赖环境变量、移除本机 FFmpeg 兜底，并在启动前拒绝无效规格及错误打斗后端；这不是新版动作能力。缺少依赖或实际可播放视频时仍须报告未完成。
用户只需提供已经写好的视频提示词。第一交付层必须是用户可直接播放的 MP4；不要先让用户操作Blender、查看编辑器、填写坐标或审查JSON。
输出合同
每次执行至少产生：
- 可播放的3D白模
MP4； previs_compiled.json：提示词的可审计3D执行解释；previs_validation.json：时长、帧数、分辨率、切点、实际解码和观看检查；- 可选
.blend工程，只作为可编辑后台产物，不代替视频。
截图、编辑器界面、工程文件、参数表、静态关键帧、2.5D推拉、软件功能说明或内存编码探针均不能替代MP4。
执行流程
- 完整保留用户原提示词。输入可为一句提示词、单镜、多镜、任意时长或无明确时长。
- 按 references/prompt-to-previs.md 编译为
previs-compiler/1.0合同。只执行提示词中已有的镜头意图；默认值必须可追踪，不得冒充正式视频模型的预测。 - 运行
scripts/validate_compiled_previs.py。时间重叠、镜头空档、缺摄影机、人物引用错误或不可执行关键歧义必须在渲染前拒绝。 - 普通走位与摄影机预演用
scripts/run_blender_previs.py调用scripts/blender_previs_adapter.py。包含合格previs-action/1.0动作扩展的打斗规格改用scripts/run_blender_rigged_fight.py调用scripts/blender_rigged_fight_adapter.py；该后端逐帧求解肩肘腕、髋膝踝、双手握点、武器和共享接触。动作后端失败时必须停止，不能暗中回退到普通体块路径。 - 用实际解码器核对MP4的帧数、帧率、分辨率和时长；再检查每镜首帧、中点、尾帧及切镜前后。
- 第一回复直接展示或链接MP4，再说明默认值和边界。缺少可播放视频时，只能报告未完成。
编译边界
- 明确时长按原文执行；未写时长时，单镜默认4秒，多镜默认每镜3秒，并标记
deterministic_default。 - 只有镜号、时间码、硬切、切到或明确转场才自动拆镜。“随后、然后、接着”默认只是同镜动作先后。
- 未写摄影机运动时保持静止，不自动添加电影感漂移。
- 未写场景时只生成地面网格和必要空间参照，不编造环境美术。
- 当前公开包已实现人形／人形机甲和双臂悬浮类代理。四足、车辆或其他未实现类别须在编译阶段报出能力缺口，不承诺已有对应后端，也不得改成人形或球体冒充原主体。除非提示词主体就是球体，不得用球替代人形主体。
- 人物和场景可以粗糙，火柴人级精度允许；但请求了运镜时，前中后景必须提供足够参照，让视差和摄影机运动在视频中肉眼可见。
- 预演证明的是当前编译解释中的镜头、空间、走位、切镜和节奏，不证明正式模型会生成相同像素、表演、材质或特效。
What ships with it
11 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.
- agents/openai.yaml 326 B
- assets/sample-previs.json 2.9 KB
- references/previs-spec.md 4.0 KB
- references/prompt-to-previs.md 6.9 KB
- scripts/blender_previs_adapter.py 33 KB runs code
- scripts/blender_rigged_fight_adapter.py 89 KB runs code
- scripts/render_previs.py 32 KB runs code
- scripts/run_blender_previs.py 4.7 KB runs code
- scripts/run_blender_rigged_fight.py 4.6 KB runs code
- scripts/validate_compiled_previs.py 12 KB runs code
- scripts/validate_spec.py 9.3 KB runs code
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.
- 4d ago First seen · 78 lines · 117 tokens per session scan A 11b1c5d9b428
whitebox-previs-executor is a skill published in the GitHub repository 62656456/ai-film-skills (17 stars, last pushed yesterday), licensed Apache-2.0. It adds 117 tokens to every session and 1,993 once invoked, about $0.0006 per session on Opus 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 2026-09-08.
Other skills, from other repositories
seedance-20
This skill should be used when creating, improving, or troubleshooting Seedance 2.0 video on any surface — Dreamina, Jimeng, CapCut, Doubao, Volcengine/Ark, BytePlus, Runway's Seedance route, fal, or third-party provider/router surfaces such as EvoLink, OpenRouter, Kie.ai, PiAPI, LaoZhang, Runware, ModelsLab, AI/ML…
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.
script-to-shootable-storyboard
A guide for turning scripts, story outlines, dialogue, or existing storyboards into detailed shot plans. A storyboard is a sequence of planned camera shots showing what viewers see and hear.
travel-skill
A production guide for realistic travel-promotion videos that combine authorised real footage with AI-generated shots.
minimax-h3
Write, debug and structure prompts for MiniMax H3 video generation (T2VA, I2VA, FL2VA, L2VA, Ref2VA) and configure its ComfyUI workflow. Use when the user mentions MiniMax H3, minimaxh3, fl2va, ref2va, MiniMaxH3ReferenceToVideo, reference-to-video, asks to animate a photo, write a video prompt, pick a model quant, or…
minimax-h3-prompting
Guide an idea into a generation-ready MiniMax H3 / Hailuo H3 prompt or diagnose an inspected H3 output. Use for T2VA, I2VA, FL2VA, L2VA, or Ref2VA; text-only PV and kinetic type, Motion Design/MG, packaging and transitions, product/UI/game/MV/title work, localized reality-to-hand-drawn edits, audio/timbre reference…