Video StudioSAFE
Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.
Overview
Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.
5ff3ca429e5bOBSERVED · 2026-10-08Host compatibility
What the documentation claims. We have not run a compatibility test.
| Host | Status | Notes |
|---|---|---|
| codex | mentioned |
What it tells the agent
The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.
--- name: "video-studio" description: "Use when the user asks to create or edit videos end-to-end (script→video, auto-cut/jumpcut, captions/subtitles, polishing for Shorts/Reels/TikTok). Current implemented backend: local FFmpeg (probe/render/jumpcut/burn-subtitles/polish). Planned/optional backends: Remotion (motion graphics templates), VectCutAPI (CapCut/剪映 timeline editing), and video-audio-mcp (MCP tool wrapper) when available. Produces a finished video artifact (MP4 by default) from assets + copy + a design/storyboard plan." --- # Video Studio (自动视频制作 / 剪辑 / 打磨) 目标:把「素材(视频/音频/图片)+ 文案 + 编排/设计方案」自动变成可发布的成片,并在最后做一轮“打磨”(节奏、转场、字幕、音量、编码)。 ## 何时使用 - 你要“从 0 生成一个视频”(口播/科普/产品介绍/短视频) - 你要“自动剪辑已有视频/音频”(去静音、裁切、拼接、转场、加字幕、加 Logo/BGM) - 你要“适配平台规格”(9:16 / 1:1 / 16:9、码率、分辨率、时长上限) - 你要“自动抛光”(音量标准化、fade in/out、H.264/yuv420p、帧率、字幕烧录) ## 背后能力(融合方案) 本 skill 不绑定单一引擎,而是按可用性选择 backend: 1. **FFmpeg(基线后端,必备)** - 本地快速剪辑、拼接、字幕烧录、音频混音、编码打磨 2. **Remotion(可选,做“像设计稿一样”的动效视频)** - 程序化动效、模板化片头片尾、动态字幕/图表 3. **VectCutAPI(可选,高阶时间线后端)** - 如果你有剪映/CapCut + VectCutAPI 服务,可用 API/MCP 进行更复杂的轨道/特效/关键帧 4. **video-audio-mcp(可选,MCP 封装的 FFmpeg 工具箱)** - 如果已配置该 MCP server,可把常用 FFmpeg 操作当成工具调用(trim/concat/subtitles/overlays/remove_silence) ## 先跑一次环境探测(强烈推荐) 在当前项目目录里执行: ```powershell python C:\Users\羽裳\.codex\skills\video-studio\scripts\video_studio.py probe ``` 它会输出 JSON,说明: - `ffmpeg/ffprobe` 是否可用(没有则给出安装建议) - `node/npm/npx` 是否可用(Remotion 需要) - `vectcut_base_url` 是否可连(可选) ## 产物约定(建议) - 中间文件:`outputs/video-studio/tmp/` - 最终成片:`outputs/video-studio/final/` 你可以把素材放在: - `assets/video/` `assets/audio/` `assets/images/` `assets/fonts/` ## 最小闭环:用 spec 直接生成视频(FFmpeg backend) 1) 写一个 spec(可参考本 skill 的 `references/sample_spec.json`) 2) 运行: ```powershell python C:\Users\羽裳\.codex\skills\video-studio\scripts\video_studio.py render --spec C:\path\to\spec.json ``` ## “打磨”步骤(推荐作为最后一步) ```powershell python C:\Users\羽裳\.codex\skills\video-studio\scripts\video_studio.py polish --in .\outputs\video-studio\final\video.mp4 --out .\outputs\video-studio\final\video.polished.mp4 ``` 打磨会尽量做到: - H.264 + yuv420p(平台兼容) - 统一帧率/分辨率 - 音量标准化(EBU R128 loudnorm) - 可选 fade in/out ## 自动剪辑:JumpCut(去静音/去停顿) 适合“口播/访谈/课程录屏”这类有明显停顿的视频:自动检测静音区间并拼接,提高节奏流畅度。 ```powershell python C:\Users\羽裳\.codex\skills\video-studio\scripts\video_studio.py jumpcut --in .\assets\video\talking.mp4 --out .\outputs\video-studio\final\talking.jumpcut.mp4 ``` 常用参数: - `--silence-threshold-db`:静音阈值(例如 `-35`) - `--min-silence`:最短静音时长(例如 `0.4`) - `--pad`:保留在静音边界的 padding(避免“咬字被切掉”) ## 字幕烧录(SRT → MP4) 如果你已经有 `captions.srt`(来自转录或人工),可直接烧录: ```powershell python C:\Users\羽裳\.codex\skills\video-studio\scripts\video_studio.py burn-subtitles --in .\outputs\video-studio\final\video.mp4 --srt .\assets\captions.srt --out .\outputs\video-studio\final\video.subbed.mp4 ``` ## 需要 AI 生成素材时(可选) 这个 skill **优先复用既有 skill**,避免重复造轮子: - 图片/背景:`$imagegen`(需要 `OPENAI_API_KEY`) - 配音:`$speech`(需要 `OPENAI_API_KEY`) - 转录:`$transcribe`(需要 `OPENAI_API_KEY`) 典型全自动流程: 1) 文案 → 生成配音(speech) 2) 配音 → 生成字幕(transcribe 或手工脚本) 3) 文案/分镜 → 生成背景图(imagegen) 4) 用本 skill 的 `render/polish` 合成出片 ## 质量门禁(交付前自检) - 成片可播放(本地播放器 + 秒开) - 分辨率/帧率符合平台(例如 1080x1920@30fps) - 音量不过爆、不太小(主讲语音清晰,BGM 不盖人声) - 字幕不超出安全区(手机端不被 UI 遮挡) ## 边界 - 本 skill 会生成/编辑视频文件,但不会替你发布到平台。 - 高级“像剪映一样”的轨道编辑:优先走 VectCutAPI(如果你有服务),否则用 FFmpeg 的能力做到“够用的自动剪辑”。
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- declared (3 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (0)
No findings outside the package's declared scope.
Gates applied: no_behavioural_pass.
5ff3ca429e5bfull audit observations/trust-audit/skill/foryourhealth111-pixel__video-studio.json · Report an issue / request a re-scanAudit history
Every audit this skill has had.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | 5ff3ca429e5b | SAFE | B | 89 | first audit |
Questions
What does the Video Studio skill do?
Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.
Is Video Studio safe to install?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.
What can Video Studio access on my machine?
The audit observed that it reaches the network. Each of those is consistent with what it says it does. Secrets in the source: none found.
Which assistants does Video Studio work with?
Its documentation mentions codex. That is what the text claims, not a compatibility test we ran.
How current is this page?
The grade is for one exact copy of the source (5ff3ca429e5b), read on 2026-10-08. The repository is watched, and a new audit runs when it changes — this is the first audit.