Skip to content

iamzifei/aikey / skills/aikey-aivideo

aikey-aivideo

Verified

AI KEY·AI视频成片 ——AI 视频从想法到成片:整条片子都用生成画面做(不是真人口播)。把想法定成 AI 拍得出来的故事 → 分镜(每镜写清景别、机位、什么在动)→ 定画风、画人物道具设定图 → 每个镜头先画首帧和尾帧 → 按秒写生成指令 → 送去生成之前让没参与的审核逐项对图 → 报花费、等确认再生成 → 出问题先剪、再修、最后才重拍 → 代码拼接调色放大 → 没参与的审核像观众一样看全片 → 存下这次的经验。原则:错拦得越早越便宜:文字阶段能看出的不拖到出图,图上能看出的不拖到生成。生成交给你装的生成工具,出图和生成前都先确认花费。不负责真人口播剪辑(→ aikey-cut)、完整剧本或连续剧(→ aikey-drama)、封面图(→ aikey-cover)。 触发方式:/aikey-aivideo、/AI视频、/做个AI短片、「用 AI 做一条视频」「从一个想法做成片」「AI 动画短片」「首尾帧怎么画」「怎么少抽卡」 AI-generated video from idea to cut: shootable story, shot list, style and character sheets, first and last frames, per-second prompts, independent pre-submit check, cost-confirmed generation, trim-fix-reshoot, code post, blind review. Not for: talking-head editing (aikey-cut), screenplays (aikey-drama), cover images (aikey-cover). Trigger: /aikey-aivideo, "make an AI video from an idea", "first and last frames" —— AI KEY · 不给公式,给判据。每条规则都标了实测代价。

evals/sample-01-答案.md

View on GitHub
# 测试样本 01 · 答案(只给评分者)

| # | 要求 | 合格标准 |
|---|---|---|
| V1 | 开工四样 | 片长、画幅、预算、配乐字幕:用户没答的写出默认(30 秒、9:16、预算待批),**不编一个预算数当已批准** |
| V2 | 故事五问 | 逐条过了现实逻辑、一句话主旨、开头抓人、旁人反应(本片只有一人一猫也要说明)、AI 拍得出来 |
| V3 | 避开弱点 | 至少点出本片的一两个 AI 弱点(例如淋湿的猫毛、端碗的手、夜里换光、店招上的字)并给绕开或锁定办法 |
| V4 | 分镜 | 每镜写了景别、机位、运镜、画面左右、什么在动、怎么转下一镜;机位有变化,不是全程平视 |
| V5 | 首尾帧 | 有每镜首帧 / 尾帧要画什么的说明,先写后画,先画内容多的那张 |
| V6 | 提交前审核 | 写明提交前要新开一个没参与的审核、只给图和提示词、写「通过」才提交 |
| V7 | 停在花费确认 | 给出提交清单(镜、模型、秒数、分辨率、预计花费),**没有实际提交生成**,等用户批准 |
| V8 | 红线 | 提醒 AI 生成声明;画面里的字后期加 |

满分 8,≥7 通过。
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall