anbeime/skill / skills/anything2explainer/anything2explainer
anything2explainer
Unverified给一个主题,产出一条黑底 MG 风格(幕底可选星点或点阵波)、有配音字幕章节进度条的科普讲解视频(中文或英文;Remotion 代码动画;时长由用户定,常用 3–5 分钟)。内含可编译模板、图元库、配音/分镜/渲染工具、风格与动效规范、多 agent 分工协议与 QC 判据,以及一条完整样片(《RAG 与知识库》)作为质量标尺。Turn any topic into a narrated motion-graphics explainer video in Chinese or English, on a black canvas with a star-field or dot-field backdrop, TTS voiceover, subtitles and a chapter progress bar, every frame drawn in code with Remotion. Use when the user asks for an explainer, educational or science-communication video about a topic, or wants an article or document turned into a video.
examples/rag/shots_src/G6/SC29.tsx
View on GitHubimport React from 'react';
import {useCurrentFrame} from 'remotion';
import {easeInOutPow, emphasisPulse, clamp01} from '../../common';
import {Box, Pill, TechText, Svg, LineArrow, LLMIcon, fadeIn, slideUp, scaleIn, exitAccel, abs} from '../../ui';
import {AnswerCard, EvidenceCard, TypeLine, HLine, typedCount, citeOp, ANS_TOTAL} from './ui';
/**
* SC29 生成:约束提示词(5314–5564)
* 节拍:5322 最后生成 / 5342 问题和证据交给大模型 / 5407 只根据资料回答 / 5472 答不出来就说不知道 / 5517 标注来源
* 画面:左「Prompt」卡(问句 + 证据 [1][2][3] + 三条规则打字机),中 LLMIcon「大模型」,右 答复卡(打字机 + 角标 8 帧淡入;闪烁整改:SC29 不在白名单,角标不再 glitch)。
* 5342 问句与证据的副本沿箭头飞入大模型;末 14 帧 Prompt 卡 / 大模型 / 箭头加速下摇淡出,答复卡留给 SC30 承接。
*/
const F0 = 5314;
const B_GEN = 5322, B_FEED = 5342, B_RULE1 = 5407, B_RULE2 = 5472, B_RULE3 = 5517;
const T_EXIT = 5551;
const PCARD = {x: 110, y: 195, w: 380, h: 410};
const LLM = {cx: 620, cy: 400, size: 130};
export const ANS_SC29 = {x: 740, y: 310}; // 答复卡 420×180,中心 (950,400);SC30 从此位置承接
const Q_PILL = {x: 150, y: 228, w: 300, h: 38};
const EVI_X = [150, 262, 374], EVI_Y = 285, EVI_W = 96, EVI_H = 62;
const RULES = ['① 只根据以下资料回答', '② 资料不足就回答:不知道', '③ 每句结论标注来源 [n]'];
const RULE_Y = [392, 452, 512];
const RULE_T0 = [B_RULE1, B_RULE2, B_RULE3];
const easeFly = easeInOutPow(2.5);
export const SC29: React.FC = () => {
const N = useCurrentFrame() + F0;
// ---- 离场(Prompt 卡 / LLM / 箭头)----
const ne = N - T_EXIT;
const exitDy = exitAccel(ne, 0.9);
const exitOp = ne <= 0 ? 1 : Math.max(0, 1 - ne / 13);
// ---- Prompt 卡入场 ----
const np = N - (B_GEN - 4);
const pDy = slideUp(np, 40);
const pOp = fadeIn(np, 10);
// ---- LLM ----
const nl = N - B_GEN;
const llmS = scaleIn(nl) * emphasisPulse(N - (B_FEED + 26), {peak: 1.12});
const arrow1 = clamp01((N - (B_FEED - 2)) / 12);
const arrow2 = clamp01((N - (B_RULE1 - 11)) / 12);
// ---- 飞入副本:问句 + 3 张证据 ----
const flyItems = [
{cx: Q_PILL.x + Q_PILL.w / 2, cy: Q_PILL.y + Q_PILL.h / 2, w: Q_PILL.w, h: Q_PILL.h, kind: 'q' as const},
...EVI_X.map((x, i) => ({cx: x + EVI_W / 2, cy: EVI_Y + EVI_H / 2, w: EVI_W, h: EVI_H, kind: i as 0 | 1 | 2})),
];
const flies = flyItems.map((it, i) => {
const t = N - (B_FEED + 3 * i);
if (t < 0 || t > 26) return null;
const u = easeFly(clamp01(t / 24));
const cx = it.cx + (LLM.cx - it.cx) * u, cy = it.cy + (LLM.cy - it.cy) * u;
const s = 1 - 0.72 * u;
const op = 1 - clamp01((u - 0.72) / 0.28);
return (
<div key={i} style={{...abs(cx - it.w / 2, cy - it.h / 2, it.w, it.h), transform: `scale(${s.toFixed(3)})`, transformOrigin: '50% 50%', opacity: op}}>
{it.kind === 'q' ? <Pill x={0} y={0} w={it.w} h={it.h} text="差旅报销的上限是多少?" fontSize={22} weight={600} sw={2} /> : <EvidenceCard x={0} y={0} w={it.w} h={it.h} label={`[${it.kind + 1}]`} seed={it.kind + 3} />}
</div>
);
});
// ---- 答复卡 ----
const na = N - (B_RULE1 - 6);
const ansDy = slideUp(na, 120);
const ansOp = fadeIn(na, 8);
const T_TYPE = B_RULE1 + 9;
const typed = typedCount(N, T_TYPE, ANS_TOTAL, 2);
const typing = N >= T_TYPE && N < T_TYPE + 2 * ANS_TOTAL + 6;
return (
<div style={{position: 'absolute', inset: 0}}>
{/* ===== 可离场组:Prompt 卡 + LLM + 箭头 ===== */}
<div style={{position: 'absolute', inset: 0, transform: `translateY(${exitDy.toFixed(2)}px)`, opacity: exitOp}}>
{np >= 0 ? (
<div style={{position: 'absolute', inset: 0, transform: `translateY(${pDy.toFixed(2)}px)`, opacity: pOp}}>
<Box x={PCARD.x} y={PCARD.y} w={PCARD.w} h={PCARD.h} r={12} sw={2} />
{/* 标题 tab:Prompt */}
<Box x={130} y={180} w={124} h={32} r={16} sw={2} />
<TechText cx={192} cy={196} text="Prompt" fontSize={24} scaleX={0.9} weight={700} />
{/* 问句 + 证据 */}
<Pill x={Q_PILL.x} y={Q_PILL.y} w={Q_PILL.w} h={Q_PILL.h} text="差旅报销的上限是多少?" fontSize={22} weight={600} sw={2} />
{EVI_X.map((x, i) => <EvidenceCard key={i} x={x} y={EVI_Y} w={EVI_W} h={EVI_H} label={`[${i + 1}]`} seed={i + 3} />)}
<HLine x={130} y={368} w={340} />
{/* 三条规则(打字机) */}
{RULES.map((r, i) => <TypeLine key={i} x={140} y={RULE_Y[i] - 16} text={r} N={N} f0={RULE_T0[i]} />)}
</div>
) : null}
{/* LLM */}
{nl >= 0 ? (
<div style={{position: 'absolute', inset: 0, transform: `translate(${LLM.cx}px,${LLM.cy}px) scale(${llmS.toFixed(3)}) translate(${-LLM.cx}px,${-LLM.cy}px)`, transformOrigin: '0 0', opacity: fadeIn(nl, 6)}}>
<LLMIcon cx={LLM.cx} cy={LLM.cy} size={LLM.size} label="大模型" labelSize={26} />
</div>
) : null}
{/* 箭头 */}
<Svg>
<LineArrow x0={PCARD.x + PCARD.w + 6} y0={LLM.cy} x1={LLM.cx - LLM.size / 2 - 6} y1={LLM.cy} p={arrow1} headL={16} headW={18} />
<LineArrow x0={LLM.cx + LLM.size / 2 + 6} y0={LLM.cy} x1={ANS_SC29.x - 6} y1={LLM.cy} p={arrow2} headL={16} headW={18} />
</Svg>
{/* 飞入副本 */}
{flies}
</div>
{/* ===== 答复卡(不离场,SC30 承接) ===== */}
{na >= 0 ? (
<div style={{position: 'absolute', inset: 0, transform: `translateY(${ansDy.toFixed(2)}px)`, opacity: ansOp}}>
<AnswerCard x={ANS_SC29.x} y={ANS_SC29.y} typed={typed} citeShow={N >= B_RULE3} citeOp={citeOp(N, B_RULE3)} cursor={typing} />
</div>
) : null}
</div>
);
};
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 470k | Updated | Claude CodeCodex | Unverified | Install |
| 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: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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 available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |