Communitygithub.com

Akira-TL/akira-video-skills

Orchestrates whole-project editing, compositing, full-film audio, and final master export. Use when multiple approved shots need to enter post-production in Premiere Pro, After Effects, DaVinci Resolve, or similar tools, or when the project requires subtitles, logos, music, mixing, aspect-ratio versions, and final delivery.

What is akira-video-skills?

This skill is designed for AI agents handling the entire video post-production pipeline at the project level. It manages final editing, sound mixing, subtitles, logos, and delivery exports—not per-shot processing. Individual shot video, audio, dialog, and images remain in their original `video/shots/<shot-id>/` directories; cross-shot materials stay in `video/materials/`. The skill only creates a `video/edit/` directory when the project genuinely enters full post-production. It reads `EDITING-RHYTHM.md` to resolve story flow, action cuts, reaction timing, sound bridges, and product legibility before applying decorative transitions. The skill enforces a 'deterministic post' philosophy: tasks precisely achievable with post-production tools (cuts, titles, logos, audio sync, final encode) are handled there, not regenerated by generative models. It also integrates QC scripts (e.g., `delivery_qc.py` using ffprobe/FFmpeg) to verify technical specs like duration, aspect ratio, resolution, codec, and audio channels against project requirements. Supported platforms include Premiere Pro, After Effects, DaVinci Resolve, and common FFmpeg workflows. The skill avoids creating duplicate, untrackable 'final' files and ensures all upstream facts (shots, scripts) are corrected at source, not silently overridden in the edit.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/Akira-TL/akira-video-skills/tree/HEAD/skills/video/video-editing

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Video Editing

This skill handles only whole-project post-production. Individual shot video, images, dialog, or audio continue to reside in the corresponding video/shots/<shot-id>/ directories; cross-shot reusable materials remain in video/materials/.

1. Create edit on demand

The post-production directory and export naming follow references/EDIT-PROJECT.md to maintain the shallowest possible structure.

Only create video/edit/ when the project genuinely enters full post-production. The directory can directly contain:

  • Premiere Pro, After Effects, DaVinci Resolve, or other post-production projects;
  • Whole-project music, narration, or mixes;
  • Subtitles, logos, titles, or composite resources;
  • Review exports and final masters.

Do not add audio/, delivery/, or other top-level directories for organizational tidiness alone; add subdirectories only when the sheer file count makes basic navigation or readability difficult.

2. Assemble currently accepted shots

When multiple accepted shots enter a rough cut, read references/EDITING-RHYTHM.md. First resolve story/information flow, action cut points, reaction timing, sound bridges, and product legibility, then consider decorative transitions.

Bring the generated results currently accepted for each shot into the timeline. Shot files remain in their original locations as long-term sources; post-production projects may reference, proxy, or manage media as needed by the tooling, but do not copy multiple hard-to-track versions of the same source video for a "final" version. Derived results such as cuts, color grading, stabilization, or replaced audio do not create new shot identities, nor should they overwrite the original generated outputs. When the current project uses video-shot, its shot identity rules continue to be governed by that skill.

3. Deterministic post-production

Tasks that can be performed precisely by post-production tools should be handled deterministically: cuts and timeline assembly, subtitles and titles, logos and precise typography, aspect ratio adaptation, simple compositing, volume and mixing, audio-video sync, final encoding, and export. For color, exposure, stabilization, noise reduction, upscaling, and texture consistency across multiple models/batches, refer to references/FINISHING.md. Discrepancies that can be fixed deterministically should not be regenerated, but human identity, product structure, and serious timeline errors cannot be masked by finishing. Handle subtitles, logos, titles, product text, and end cards per references/GRAPHICS-TITLES.md; do not rely on a generative model to accurately render text. When the project requires dialog/narration subtitles, external subtitles, burned-in subtitles, accessibility captions, or multilingual subtitles, then read references/SUBTITLES.md, generate from the final audio/text, and check timing and safe zones for every language and aspect ratio.

Do not let a generative model redo tasks that deterministic post-production can handle reliably simply because it can attempt them.

When preparing for final export or whole-project review, first check the entire film content against references/FINAL-QC.md, then verify each official delivery file against references/TECHNICAL-QC.md for proper decoding, duration, aspect ratio, resolution, audio tracks, and other explicit delivery requirements. If ffprobe/FFmpeg is available in the current environment, prioritize using this skill's bundled scripts/delivery_qc.py to perform complete decoding and explicit parameter validation on official files; check width, height, aspect ratio, duration, FPS, container, video/audio codec, audio tracks, channels, sample rate, and square pixels as needed. Expected values must come from VIDEO.md / client / competition / platform requirements; the script does not provide generic delivery defaults and does not replace a full content review.

4. Full-film inspection

Review the final master from start to finish for plot/information continuity, shot transitions, music and dialog, subtitles/logos, product model numbers, or any other delivery requirements. Whole-project issues must be corrected by returning to the corresponding shot, material, or script; do not silently override upstream facts within the post-production project.

Completion criteria

The whole-project edit can clearly identify the currently accepted shots and all project-level resources. The final master satisfies the current delivery requirements specified in VIDEO.md, and no duplicate, untrackable "final" files have been created.

Individual skills in this repo

This repo contains 12 individual skills — each has its own dedicated page.

Akira-TL/akira-video-skills

为 Runway 当前视频模型编译和检查生成提示词;当 video-generation 需要使用 Runway(当前主线 Gen-4.5)的 Text to Video 或 Image to Video,并需要 Runway 专用动作、摄影和时序表达时使用。

Akira-TL/akira-video-skills

为 ByteDance / BytePlus Seedance 当前视频模型编译和检查提示词;当 video-generation 需要使用 Seedance 2.5 的参考素材、时间线、同步声音、视频编辑、延长或多镜头能力时使用。

Akira-TL/akira-video-skills

为 Google Veo 视频生成编译和检查提示词;当 video-generation 需要使用 Veo 的 Text to Video、Image to Video、首尾帧、参考元素或同步声音能力时使用,并先核验具体 Google 使用入口与 model 的当前支持范围。

Akira-TL/akira-video-skills

进入并持续推进一个 AI 视频制作项目;读取或建立 VIDEO.md,根据当前视频脚本、复用素材、镜头、生成结果和整片后期状态,按需路由到 canonical 视频专业 Skill,而不是把所有制作规则塞进一个总提示词。

Akira-TL/akira-video-skills

为品牌短片、产品植入剧情、广告或比赛作品建立产品事实与广告边界;当视频需要准确表达某个具体产品的官方功能、参数、结构、使用方式、品牌素材或产品与剧情关系时使用。

Akira-TL/akira-video-skills

设计并维护 AI 视频中跨镜头复用的角色声音、音色参考、对白声音要求与整片声音关系;当同一角色需要跨镜头保持声音一致,或需要规划语音参考、环境声、音效、音乐与画面的职责边界时使用。

Akira-TL/akira-video-skills

设计 AI 视频中的角色与场景视觉方案;当脚本只说明“需要一个什么样的人/场景”,但还没有形成可稳定生成的外形、服装、材质、色彩、空间、光线和视觉识别系统时使用。

Akira-TL/akira-video-skills

为 AI 图片或视频生成准备可直接使用的提示词与一次性生成包,并在用户把生成结果带回项目后归档;当需要从正式视频脚本、复用素材和 SHOT.md 编译生成内容、选择参考素材、约定返回文件名或清理临时包时使用。

Akira-TL/akira-video-skills

规划、创建、整理和维护多个镜头会重复使用的视频复用素材;当需要人物/场景/道具/产品提示词、参考图、官方素材、身份与状态参考或其他可重复引用内容时使用。

Akira-TL/akira-video-skills

审核 AI 生成的人物/场景等复用素材或镜头生成结果,判断采用、修改提示词、重生成还是留给后期修复;当生成结果返回项目、需要检查身份/结构/连续性/动作/摄影/产品真实性或选择生成结果时使用。

Akira-TL/akira-video-skills

把小说、章节、故事梗概、品牌 brief 或其他上游内容转换成当前视频自己的脚本层;当需要编写或修订 SCRIPT.md、视频人物表演定义、场景定义、对白或镜头总表时使用。

Akira-TL/akira-video-skills

把 SHOTS.md 中的镜头总览展开成可直接制作的单镜头定义;当需要创建或修改 video/shots/<shot-id>/SHOT.md、镜头连续性、摄影、动作、表演、声音或镜头专属生产约束时使用。

Related Skills