{"slug":"shortfilm-prompt","title":"shortfilm-prompt","summary":"Generate cinematic AI shortfilm prompts (works with Seedance 2.0, Xiaoyunque, Sora, Kling, Jimeng, Veo) using the 5-stage structure from Mx-Shell's Zombie Scavenger. Trigger when the user wants transformation sequences, multi-shot narrative shorts, weapon-charge/combat segments, ","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-08-25T15:08:20.563801Z","repo":{"url":"https://github.com/jnMetaCode/ai-shortfilm-prompts","stars":446,"forks":86,"license":"MIT","updatedAt":"2026-09-28T06:36:40Z"},"bodyHtml":"<h1>shortfilm-prompt 测试集</h1>\n<blockquote>\n<p>用来验证 SKILL.md 真的在生效。\n每个 case 都是「用户输入 + 期望输出特征 + 容易翻车的点」。</p>\n</blockquote>\n<h2>怎么用这个测试集</h2>\n<h3>方法 1：手动验证</h3>\n<ol>\n<li>打开新的 Claude Code 窗口</li>\n<li><strong>不要</strong> 加载 shortfilm-prompt skill</li>\n<li>把 <code>01-mecha-energy-shield.md</code> 里的用户输入复制给 Claude</li>\n<li>观察输出 —— 大概率会犯 7 条规则里的几条（这是基线，记录下来）</li>\n<li>加载 skill 后再跑一次同样的输入</li>\n<li>对比：7 条规则是否全过</li>\n</ol>\n<h3>方法 2：自动验证（如果你想做 CI）</h3>\n<p>在 SKILL.md 里加 evaluation 段，用 LLM-as-judge 跑每个 example，看输出是否命中所有期望特征。</p>\n<hr>\n<h2>测试 Case 列表</h2>\n<p>共 5 个 case，分布在 6 个文件里（Case 01 含「用户输入」与「Skill 输出范例」两份）：</p>\n<table>\n<thead>\n<tr>\n<th>文件</th>\n<th>Case</th>\n<th>检验目标</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><code>01-mecha-energy-shield.md</code></td>\n<td>Case 01 · 单镜头能量护盾开启（绿色配色 + 雷暴码头）</td>\n<td>不抄原始库题材，5 段式骨架完整</td>\n</tr>\n<tr>\n<td><code>02-skill-output-sample.md</code></td>\n<td>Case 01 的 Skill 输出参考样本</td>\n<td>10 条 checklist 全过的范例输出</td>\n</tr>\n<tr>\n<td><code>03-multi-shot-cat-encounter.md</code></td>\n<td>Case 02 · 多分镜叙事（机器人 + 流浪猫相遇）</td>\n<td>切换到写法 B（按镜头切片），不死守按秒切片</td>\n</tr>\n<tr>\n<td><code>04-weapon-charge-combat.md</code></td>\n<td>Case 03 · 武器充能 + 打斗（强制两段独立）</td>\n<td>识别经典分段套路，建议分两段生成 + 后期剪辑</td>\n</tr>\n<tr>\n<td><code>05-ip-name-forced.md</code></td>\n<td>Case 04 · 用户强制要求用 IP 名</td>\n<td>照写但末尾加拦截提示 + 给替换方案</td>\n</tr>\n<tr>\n<td><code>06-emotional-pet-farewell.md</code></td>\n<td>Case 05 · 情感叙事（萌宠亲情 · 一生陪伴）</td>\n<td>识别情感叙事类型 + 加载亲情模板，时间锁调色、结尾克制、瑕疵=一致性锁</td>\n</tr>\n</tbody>\n</table>\n<hr>\n<h2>评分标准（10 条 checklist）</h2>\n<p>跑完输出后用这个核对：</p>\n<ul>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 5 段结构齐全</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 摄影机型号 + 镜头型号</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> \"如呼吸般的镜头浮动\" 完整句式</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> \"声音：不需要配乐，仅保留同期声\"</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> ≥2 处瑕疵描述</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 结尾不堆特效（不出现 \"光芒万丈/爆炸/胜利姿态\"）</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 无空泛词（不出现 \"完美/震撼/史诗/帅气/4K/质感拉满\"）</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 无 IP 名 OR 有 IP 名时末尾有拦截提示</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 单镜头≤15s / 多镜头≤8 分镜</li>\n<li><input disabled=\"disabled\" type=\"checkbox\"> 末尾给目标模型兼容性建议</li>\n</ul>\n<p>≥9 条通过 = skill 合规。</p>\n","files":[{"path":"examples/01-mecha-energy-shield.md","sizeBytes":2060,"isText":true},{"path":"examples/02-skill-output-sample.md","sizeBytes":6365,"isText":true},{"path":"examples/03-multi-shot-cat-encounter.md","sizeBytes":2137,"isText":true},{"path":"examples/04-weapon-charge-combat.md","sizeBytes":2487,"isText":true},{"path":"examples/05-ip-name-forced.md","sizeBytes":1397,"isText":true},{"path":"examples/06-emotional-pet-farewell.md","sizeBytes":2652,"isText":true},{"path":"examples/README.md","sizeBytes":2411,"isText":true},{"path":"SKILL.md","sizeBytes":19421,"isText":true},{"path":"SKILL.zh.md","sizeBytes":18290,"isText":true},{"path":"TESTING.md","sizeBytes":4652,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-08-25T15:08:51.918925Z","sha256":"51659025B55C350773849E7A98F60AB5438CD7DD6CF03B1AB743AB805D1830B8","sizeBytes":33482},"review":null,"source":{"repositoryUrl":"https://github.com/jnMetaCode/ai-shortfilm-prompts","path":"skills/shortfilm-prompt","license":"MIT","commit":"9c4536ed1a885f12f9da9c20bbc475194876f14c","subtreeSha":"ADBD30977AB9FD85E0660C9CD1031C76047D20C2C4407CB6C539A05C9E22DAF0","lastSyncedAt":"2026-09-29T23:33:19.425354Z"},"reviewedAt":"2026-08-25T15:10:08.605116Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/jnMetaCode/ai-shortfilm-prompts/tree/main/skills/shortfilm-prompt"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install jnmetacode-ai-shortfilm-prompts@llmmart"},{"target":"git","command":"git clone https://github.com/jnMetaCode/ai-shortfilm-prompts.git"}]}