Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add cyuanxv/ai-mandrama-skills --skill ai-mandrama-pipelinegit clone --depth 1 https://github.com/cyuanxv/ai-mandrama-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cyuanxv/ai-mandrama-skills/ai-mandrama-pipeline)<a href="https://agentmods.dev/skills/cyuanxv/ai-mandrama-skills/ai-mandrama-pipeline"><img src="https://agentmods.dev/badge/skills/cyuanxv/ai-mandrama-skills/ai-mandrama-pipeline/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/cyuanxv/ai-mandrama-skills/ai-mandrama-pipeline"><img src="https://agentmods.dev/badge/skills/cyuanxv/ai-mandrama-skills/ai-mandrama-pipeline.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.01121 | $0.05934 |
| Opus 5 | $0.00561 | $0.02967 |
| Sonnet 5 | $0.00224 | $0.01187 |
| Haiku 4.5 | $0.00112 | $0.00593 |
Grade A, and why
ai-mandrama-pipeline scanned grade A with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
urllib.request.urlretrieve(url_m.group(1), out_path) Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
r = subprocess.run([ How it starts
The opening of the file, as written. The whole thing — 314 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI 漫剧端到端制作流水线（中文）
何时使用本 Skill
- 用户有剧本一卡 + 分镜表，要把它做成中文 AI 漫剧/短剧
- 用户在跑"鱼子酱 AI 漫剧"课程 SOP 实操，要 Dreamina + ffmpeg 全套
- 用户问"分镜→生图→生视频→配音→拼接"任意环节的踩坑/最佳实践
- 用户要做 1-3 分钟横版 + 9:16 竖版双端成片
- 用户睡觉前要"无人值守自动跑完一整集"
- 用户用 dreamina 视频任务遇到队列/并发限制
- 用户用 ffmpeg drawtext 烧中文字幕遇到方块/转义问题
核心流程（SOP 9 阶段）
阶段 1-3：素材准备
| 阶段 | 文件 | 说明 |
|---|---|---|
| 1. 剧本一卡 | 00_剧本/一卡_v1.docx |
动态漫剧本，每镜头描述场景+人物+动作+对白 |
| 2. 分镜表 | 01_分镜表/第一集_分镜表_v1.xlsx |
9 列结构（镜头号/场景/人物/动作/提示词/运镜/对白/对应对白/备注） |
| 3. 角色介绍 | 02_核心场景与角色/第一集_核心场景与角色.docx |
角色设定参考（外貌/服装/性格） |
项目目录强制规范：
~/Downloads/AI短剧_<剧名>/
├── 00_剧本/
├── 01_分镜表/
├── 02_核心场景与角色/
├── 03_资产库/第一集/角色/ ← 三视图存这
├── 04_单镜头生图/第一集/ ← 14 张分镜单图
├── 05_单镜头生视频/第一集/ ← 13-14 段 mp4
├── 06_首尾帧素材/第一集/ ← 可选
├── 08_成片/ ← 最终交付
└── README_制作过程.md
阶段 4：角色资产库（三视图）— 关键迭代
风格选型决策：
- ❌ 写实风（"8K高清 + 电影级质感 + 写实风格"）→ 容易让爽文角色看起来太"棚拍模特"
- ✅ 日漫赛璐璐二次元（"日漫赛璐璐平涂上色，类似《全职高手》《魔道祖师》《天官赐福》乙游风格"）→ 漫剧标配
- ⚠️ 治愈系日漫（《夏目友人帐》风）→ 太软，爽文不适用
用户反馈"明媚温柔"≠ 切换风格基调，而是相对方向微调——保持二次元基调 + prompt 加入"温柔/明亮/带笑意"。经验教训：不要矫枉过正。
角色 prompt 速查（详细模板见 reference.md）：
| 角色类型 | 关键词 |
|---|---|
| 爽文男主 | 桃花眼略上挑、剑眉斜飞、薄唇微抿带邪魅、黑色微长碎发气流感、慵懒锋利眼神 |
| F 级女主/可怜系 | 纤瘦娇小、黑色长发披肩微卷、樱桃唇、楚楚可怜被欺负、破旧但本身好看的反差 |
| 配角性感美女 | 沙漏型胸大腰细、浅棕色大波浪卷长发、温柔明媚大姐姐 |
| 反派/粗暴男 | 身材魁梧、满脸横肉胡渣、黑色破皮衣 |
命令模板：
dreamina text2image \
--prompt="高质量日韩漫画动漫人物设定稿，<角色>三视图：正面/侧面/背面横向并排，纯白背景，无任何水印文字。<角色五官身材服装详细描述>。日漫赛璐璐平涂上色，干净流畅线条，精致五官刻画，类似《全职高手》乙游风格。三视图保持发型五官身材服装严格一致，仅角度不同。8K高清，无文字，无水印。" \
--ratio=16:9 --resolution_type=2k --model_version=5.0 --poll=90
同角色换装（如男主庄园造型 vs 集市造型）用 image2image 拿三视图当锚定，保证人脸一致性：
dreamina image2image \
--images="$角色_庄园造型_三视图.png" \
--prompt="保持参考图人物的脸/发型/五官/身材/画风/纯白背景/三视图布局完全一致，只把服装换成<新服装>" \
--ratio=16:9 --resolution_type=2k --model_version=5.0 --poll=120
阶段 5：单镜头生图（14 张）
核心心法：有人物的镜头必须用 image2image + 角色三视图作锚定，否则角色每帧脸都不一样。
判定规则：
| 镜头特征 | 命令 | 锚定 |
|---|---|---|
| 无人物（废墟/环境/系统面板） | text2image |
— |
| 单人物近景/中景 | image2image |
该角色三视图 |
| 双人物互动 | image2image |
两个角色三视图（多图作 --images） |
| 多人物群像 | image2image |
主角三视图 + prompt 描述其他人 |
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 314 lines · 1,121 tokens per session scan A 0eb836e3d59c
ai-mandrama-pipeline is a skill published in the GitHub repository cyuanxv/ai-mandrama-skills (23 stars, last pushed 3mo ago), licensed MIT. It adds 1,121 tokens to every session and 5,934 once invoked, about $0.0056 per session on Opus 5. A static security scan graded it A with 2 findings (makes network calls, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
story-to-video-studio
Turn a story, novel excerpt, treatment, or screenplay into a complete AI-video production and editable post-production package. Use when Codex needs to adapt narrative, design characters and locations, write a timed shooting script, generate storyboard and image-to-video prompts, manage generation budgets and approval…
frontend-design
Aesthetic direction for OrkasVideoStudio HTML and motion-graphics compositions. Use before stage-compose writes composition-manifest.json artdirection and index.html to choose a subject-specific visual point of view, type, palette, layout signature, restrained motion, and anti-template checks.
stage-plan
The "ingest + plan" half of end-to-end video orchestration — ingest the user's material from evidence, then decompose intent into ONE cross-modal EDL (plan.json: edit/generate/compose/provided segments + narration/music/caption tracks + a delivery promise), validate it with ovs plan validate. Trigger when the…
gate-control
Canonical VideoStudio review authorization and state-transition policy. Use after any Gate B/C/Preview/D decision, post-gate revision, resumed approval, or exhausted visual-QA result across COMPOSE/AUTO/GENERATE/EDIT; maps explicit user authority and durable artifact state to one next action with ovs gate transition.…
stage-edit
Intelligent editing of real user-supplied footage—understand it with transcript/OCR/scene/silence/quality/vision evidence, then choose deterministic timeline operations or a constrained semantic AI edit. Trigger for repurpose, montage, cleanup, localization, narration, or local content changes.
stage-consistency
Multi-shot narrative & character consistency — a character bible with a locked front-portrait anchor, view-matched reference selection, recent-frame carry-forward, Cameo (a user photo as the lead), and global planning for long scripts/novels. Trigger on top of the generation line when the SAME character must look the…