What it does
Zero-skill cinema. Senior-director prompts on autopilot. A Claude Code Skill for high-quality AI image / video / music prompt crafting and browser-based execution across 14+ generative platforms.
- Connects Instruction files the AI reads
- Account or key Not confirmed (README mentions a key or .env file)
- Category Coding
For makers: add a badge to your README
If you made this, paste this code into your README. The badge links back to this page.
[](https://mcp.popnx.com/en/skills/hao0321-ai-media-generator)Use together with
MCP servers that connect the AI to services for the same work
Fallow
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
Desktop Commander
MCP server for terminal commands, file operations, and process management
Similar Skills
Same kind of work, ordered like the Skills list
| This one ai-media-generator | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| Maker | Hao0321 | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skills | 1 | 1 | 1 | 1 | 1 |
| GitHub stars | 282 | 468,733 | 234,332 | 150,287 | 144,566 |
| Last code change | Aug 9, 2026 | Nov 19, 2026 | Apr 20, 2026 | Nov 18, 2026 | Nov 18, 2026 |
| License | MIT | MIT | – | Apache-2.0 | MIT |
1 skills (instruction sets) by Hao0321 that an AI reads and follows · not an MCP server
How a Skill differs from MCP
MCP = the hands
Connects the AI to an outside service or data, such as reading Notion or creating an invoice. A program or server has to be running.
Skill = the manual
A text file that teaches the AI how to do a task. It connects to nothing.
Some overlap: a few Skills run scripts that call tools, and a few MCP servers only send knowledge. Start from the task you want done.
Works with
Claude CodeCodex
From the maker’s README and the folders in the repo
Install
Is it safe?
- Who made it
GitHub account Hao0321 (person), 3 years old
- Use and upkeep
★ 282 stars, few users so far · code last changed Aug 9, 2026
- Any scripts?
No, plain text files the AI reads and follows
- Uses the internet?
No scripts, so it does not go online by itself
- Changes machine-wide settings?
No hooks or writes to machine settings found
Checked automatically from the repo files at commit edc8aa5 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
First: which AI do you use?
Pick your app and this page shows only how to install it there
Option 1Upload to claude.ai (web and app)
Uploading to claude.ai would fail: the description is over 200 characters (638). Use "Let the AI install it" instead
Option 1Let Claude Code install it
Copy this sentence and paste it into a Claude Code chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Restart Claude Code, then type /ai-media-generator or ask for a matching task and Claude picks it up
Checked against official docs Nov 12, 2026 · Source: code.claude.com
For Terminal users
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Open Terminal
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
Copy the skill folder to ~/.claude/skills
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
Restart Claude Code, then type /ai-media-generator or ask for a matching task and Claude picks it up
ChatGPT desktop app
The ChatGPT desktop app has Codex built in, and Codex reads skills from ~/.agents/skills
Option 1Let ChatGPT install it
Open Codex in the ChatGPT desktop app and paste this sentence into the chat
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Option 2Put the folder there yourself
Put the ai-media-generator folder at ~/.agents/skills/ai-media-generator
~ is your home folder ($HOME)
For Terminal users
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Open Terminal
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
Copy the skill folder to ~/.agents/skills
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
After installing
- ChatGPT desktop app: open Skills in the sidebar to see that the skill is there
- To use it: in ChatGPT type @ and pick the skill · in Codex CLI or the IDE extension run /skills or type $ and the skill name
- Codex picks up newly installed skills by itself; if one does not show up, restart Codex
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
ChatGPT on the web and mobile
Skills you install yourself do not work here; only skills that come with a plugin from the directory
To use this skill with ChatGPT, use the ChatGPT desktop app as above
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
Option 1Let Codex install it
Copy this sentence and paste it into a Codex chat · Works in Codex inside the ChatGPT desktop app, Codex CLI and the IDE extension
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex picks up newly installed skills by itself; if one does not show up, restart Codex · To use it: in ChatGPT type @ and pick the skill · in Codex CLI or the IDE extension run /skills or type $ and the skill name
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
For Terminal users
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Open Terminal
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
Copy the skill folder to ~/.agents/skills
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
Restart Codex, then ask for a matching task; Codex lists the skill
Option 1Let Cursor install it
Copy this sentence and paste it into a Cursor chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
In Cursor’s Agent chat, type / and pick ai-media-generator
Checked against official docs Nov 12, 2026 · Source: cursor.com
For Terminal users
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Open Terminal
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
Copy the skill folder to ~/.cursor/skills
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
In Cursor’s Agent chat, type / and pick ai-media-generator
Option 1Let the AI install it
For other apps that read skills. The sentence gives the paths for Claude Code, Codex and Cursor so the AI picks the one for your app · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs Nov 12, 2026 · Source: code.claude.com, learn.chatgpt.com, cursor.com
For Terminal users
Other apps that support Agent Skills (such as Gemini CLI or GitHub Copilot) use different folders. Check that app’s guide, then copy the folder with the SKILL.md there.
Skill folders checked against the official Claude Code, Codex and Cursor guides on Nov 11, 2026
View the original SKILL.md
The text below comes from the maker’s repo, not from PopNx MCP
Showing the first 8,000 characters · Open the full file on GitHub
---
name: ai-media-generator
description: 為使用者產生高品質的 AI 生圖、生影片、生音樂提示詞,並在需要時透過瀏覽器自動化實際送到目標平台。涵蓋 OiiOii、Kling 3.0/O-series、Seedance 2.0/2.5、Suno v5.5、Seedream 5.0/4.0、Vidu Q3、Midjourney V8.1、Flux 1.1 Pro / Kontext、Runway Gen-4.5 / Aleph、Google Veo 3.1、Ideogram 3、Nano Banana Pro、Stable Diffusion 3.5(⚠️ OpenAI Sora 2 已於 2026-04-26 停運,API 撐到 2026-09-24,預設改推 Runway/Veo/Kling)。只要使用者提到「AI 生圖」「AI 影片」「AI 音樂」「做 MV」「做 storyboard」「寫 prompt 給 XXX」「我想用 Kling/Suno/Midjourney/Runway/Veo...」「幫我操作 OiiOii / 即夢 / 可靈」「txt2img / img2video / 文生圖 / 文生影片 / 圖生影片」「角色一致性」「多鏡頭分鏡」「運鏡」「結果有瑕疵 / 不夠精緻 / 怎麼修」,或任何跟上述平台或影像/影片/音樂生成工作流相關的任務,都要用這個 skill。即使他們沒講明平台,只要任務是要餵給某個生成模型的 prompt,就用這個 skill 幫他們選對的平台、寫對的格式。
---
# AI Media Generator
> **Codex compatibility:** References that still say “Claude in Chrome” or “Chrome MCP” describe legacy browser behavior. On Codex, load `chrome:control-chrome` and use the currently available Chrome tools; never invent or call legacy `mcp__Claude_in_Chrome__*`, `javascript_tool`, or `browser_batch` names.
幫使用者把想法變成高品質的 AI 生成內容 (圖片、影片、音樂),核心工作是 **寫對每個平台的 prompt** 以及 **必要時自動操作網站**。
**AI 短劇邊界:** 使用者要做 AI 短劇/漫劇、重生/系統/逆襲/隱藏大佬、分集連載或整季架構時,先由 [`ai-short-drama`](https://github.com/Hao0321/ai-short-drama) 建題材、Bible、pilot、分集狀態與 production pack;本 Skill 只接下游角色圖、場景圖、逐鏡影片/聲音 prompt 與模型執行,不自行重寫季弧。
## 🤖 Auto-Pilot Mode (超傻瓜一句話到成品)
**觸發:** 使用者說「幫我做 X」「做一個 X」「生個 X」「我要 X」這類命令式、且 X 含媒體類型 (圖/影片/音樂/MV/短片/海報/動畫) 或風格關鍵字 → **直接進 auto-pilot 不問**。
**流程 (見 [templates/auto-pilot.md](templates/auto-pilot.md) 完整版):**
1. **Intent Parser** — 一句話拆 9 slot (媒體/長度/畫面比/主題/風格/角色/場景/音訊/語言)
2. **Fill Defaults** — 沒講的用預設 (video 預設 10s 16:9;動漫預設 Shinkai+Ghibli;電影預設 Deakins DP…)
3. **Platform Decider** — 按 [references/selector.md](references/selector.md) + quick pick 矩陣自動選
4. **Script + Storyboard Auto-gen** — 自動寫 logline + 角色卡 + 分鏡
5. **Prompt Crafting** — 強制套語彙庫 (見下方硬規則)
6. **Preview + Go** — 30-秒 override 視窗,使用者回「go/確認/OK」才執行;無回 + 需花錢 → 停;無回 + 免費 → 30 秒後自動繼續
7. **Execution + Report** — 按 [automation/click-protocol.md](automation/click-protocol.md) + [site-profiles/](automation/site-profiles/) 協議操作
**Auto-Pilot 必停的 checkpoint (不代做):**
- ⛔ 付費 paywall / 升級提示
- ⛔ 送出前最終一次確認
- ⛔ 下載 / 分享 / 公開發布
- ⛔ 敏感內容 (真人照片疑似裸露/暴力/政治人物)
- ⛔ click-protocol.md 定義的「irreversible action」
**Auto-Pilot 禁止行為:**
- ❌ 幫 Claude 已知的事問使用者 (「你要幾秒的影片?」— 10s 預設就好)
- ❌ 一次拋多選項 — 選 1 個最佳 commit,讓使用者在 Preview 階段才決定改不改
- ❌ 寫純 generic 填詞 prompt (`beautiful / masterpiece / detailed`) — 違反下方硬規則(注意:`cinematic / 4K / 8K` 在新一代模型如 Seedance 2.0 是有效 token,看平台分流)
- ❌ 代付款
**使用者自然語言 flags 可 override 預設:** 使用者一句話可帶「用 Kling / 5 秒 / 豎屏 / 免費 / Ghibli 風 / 有對白」等自然語言 flag,auto-pilot 掃 [templates/user-flags.md](templates/user-flags.md) 對照表自動套用。使用者不懂術語也 OK — 「做個抖音」「可愛一點」「夢幻」都有對應翻譯。
**⚡ Token-Efficient Mode (大專案必讀):** 本 skill 55+ 檔 / 14000+ 行,全量讀會爆 context。Auto-Pilot 預設套 [templates/token-efficient-mode.md](templates/token-efficient-mode.md) 的 7 層策略 (lazy load / grep / 子代理 / preset 跳過 / cache polling)。**一般任務目標 ~25-40k tokens,不是 100k+**。豪華模式 (全量讀) 只在 benchmark / 學 skill / 陌生平台探索時啟用。
## 🎯 快查表:3 秒找對檔案(更直覺入口)
**先看這張表 → 對到使用者說的話 → 直接讀那個檔。** 找不到再往下走完整流程。
| 使用者說… | 讀這個檔 | 為什麼 |
|---|---|---|
| 「AI短劇/AI漫劇/爽劇/重生/系統/逆襲/隱藏大佬/分集連載」 | [`ai-short-drama`](https://github.com/Hao0321/ai-short-drama) | 先鎖故事、角色、揭露梯與 state;媒體 prompt 再回本 Skill |
| 「結果有瑕疵/爛掉/不夠精緻/閃爍/變形/塑膠感/文字怪」 | [quality-control.md](references/quality-control.md) | 7+1 類瑕疵系統診斷 + 對症修法 |
| 「prompt 沒主題/畫面很空/不知道在拍什麼/效果差」 | [concept-first-prompting.md](references/concept-first-prompting.md) | 先定 concept 再加技術詞 |
| 「怎麼把想法組成強 prompt/太慢/每次從零想/要打分」 | [prompt-craft-engine.md](references/prompt-craft-engine.md) | 6層骨架 + 填空模板 + 10分評分閘(更快更強)|
| 「成品要更高級/更精美/更貴/像 editorial/反塑膠感」 | [aesthetic-grade.md](references/aesthetic-grade.md) | 打光比/調色/材質/鏡頭 DP 級配方(更精美)|
| 「不知道用哪個模型/該用哪個平台/不熟各家」 | [model-picker.md](references/model-picker.md) | 決策樹 + 各家最強情境 |
| 「角色要一致/同一個人出現多鏡/風格鎖定」 | [asset-library.md](templates/asset-library.md) + [multimodel-video-cheatsheet.md](references/multimodel-video-cheatsheet.md) | 角色卡 + @element 身份鎖 |
| 「Seedance 2.5/30秒/社群案例/UGC/Vlog/多參考/硬切/提示詞案例」 | [seedance-2-5-community-playbook.md](references/seedance-2-5-community-playbook.md) + [seedance.md](references/seedance.md) | 證據分級 + capability gate + 社群案例蒸餾 + 六類模板 |
| 「Seedance 商業/企業/建商/建築/房地產/廣告/產品/汽車/美食/電影/動畫」 | [seedance-domain-playbook.md](references/seedance-domain-playbook.md) + [seedance.md](references/seedance.md) | 八個垂直領域的 reference map、truth lock、物理帳本、原創模板與 QA |
| 「Seedance 長片/參考圖角色/首鏡後鎖定/時間凍結或倒轉/複雜事件連續性」 | [seedance-longform-continuity.md](references/seedance-longform-continuity.md) + [seedance.md](references/seedance.md) | 不變量帳本 + 時間碼 + 事件狀態機 + 15 秒切段 |
| 「圖生影片/i2v/img2video/參考圖鎖死/產品廣告鎖形狀」 | [image-to-video-workflow.md](references/image-to-video-workflow.md) | i2v 鎖形狀 SOP(產品廣告最穩) |
| 「寫歌/純音樂」→[suno.md](references/suno.md)|「對白/音效/配樂/原生音訊」→[sound-design.md](references/sound-design.md) | 二選一 | 純音樂 vs 影片內音訊分流 |
| 「做 MV/卡點/轉場/剪接節奏」 | [music-video.md](templates/music-video.md) + [editing-transitions.md](references/editing-transitions.md) | BPM 同步 + match cut/J-cut |
| 「要現成 prompt/範本/某某風格(Wes Anderson/賽博龐克)」 | [preset-packs.md](templates/preset-packs.md) | 30+ preset 換占位符即用 |
| 「想抄某導演/DP/品牌/製作公司風格」 | [director-style-library.md](references/director-style-library.md) | 簽名 token(⚠️ Flux/Nano 洗掉導演名)|
| 「超短廣告/TikTok/10秒快剪」 | [nano-banana.md](references/nano-banana.md) | <30 token 超精簡 prompt |
| 「Wan 2.7/Kling Omni/HappyHorse/拉片復刻/多模型省成本」 | [multimodel-video-cheatsheet.md](references/multimodel-video-cheatsheet.md) | 2026 新模型招牌技巧 |
| 「幫我直接操作 OiiOii/Flow/Kling」 | [click-protocol.md](automation/click-protocol.md) + [site-profiles/](automation/site-profiles/README.md) | 自動化協議 + 各站 UI 地圖 |
| 「免費/不想花錢」 | [selector.md](references/selector.md) Cost 欄 + 各平台檔 Free tier | 免費額度速查 |
> **🔴 OiiOii 專用:** OiiOii **2026-06-08 大改版**,最新 UI/流程/成本/i2v/選模型/新模型(Oii Image 2 / Gemini Omni / Oii X Imagine)一律以 [`automation/site-profiles/oiioii.md §0`](automation/site-profiles/oiioii.md) 為準(**SOURCE OF TRUTH**)。`references/oiioii.md` 只看 prompt 哲學,UI 別照它。
## 🔴 硬規則 (Mandatory)
### Meta 優先序(2026-04-21 實戰鐵律)
**Prompt 寫對一次 ≫ 操作快 10 次。**
根因:寫錯 prompt → 重做 → 等 8-10 分鐘。操作 25 秒 vs 5 分鐘差距(~4.5 分鐘),**遠遠小於「一次 prompt 失敗」成本(~10 分鐘等待 + token 浪費)**。所以速度優化的**真正**優先序:
1. **第一優先:查平台簽名 + 寫對 prompt** → [references/community-prompt-patterns.md](references/community-prompt-patterns.md)(跨 X/Threads/Reddit/小紅書/Bilibili 社群驗證,單一 source of truth)
2. **第二優先:單次提交極速化** → [automation/site-profiles/](automation/site-profiles/)
3. **第三優先:等待不 polling** → `Bash run_in_background:true + sleep 400`
順序反了 = 浪費 40+ 分鐘做 4-5 次嘗試才生出可用的。
### Prompt 必備語彙
**每次產 image / video / music prompt,都必須從 skill 進階語彙庫挑 token — 不是選配,是必做。**
**流程:**
1. **最優先查**:[references/community-prompt-patterns.md](references/community-prompt-patterns.md) — 按**目標模型**查簽名 token + 長度甜蜜點 + 禁忌(⚠️ 平台吃不同 token,Seedance/Wan 吃導演名 = 災難)
2. 決定任務類型 (攝影感/電影感/廣告/時尚/MV/VFX/社群短片)
3. **挑對應 reference 檔讀:**
- 電影/攝影類 → [cinematic-direction.md](references/cinematic-direction.md) 必讀
- 廣告/時尚/MV/品牌 → [commercial-direction.md](references/commercial-direction.md)
- 特效/物理/大氣 → [vfx-effects.md](references/vfx-effects.md)
- 原生音訊 (Veo/Sora/Vidu Q3) → [sound-design.md](references/sound-design.md)
- 多鏡/剪接/節奏 → [editing-transitions.md](references/editing-transitions.md)
4. **快速路徑**:先查 [preset-packs.md](templates/preset-packs.md) 找最近的 preset,換占位符即可
5. **每個 prompt 至少嵌入 5-8 個高訊號 token**,從下列層挑(**看平台分流**):
- 導演/DP 名 (Deakins、Lubezki、Hoytema、王家衛、新海誠…) — ✅ MJ/Sora 2/Veo;❌ Flux/Nano Banana Pro/Seedance/Wan
- 鏡頭/焦段 (Panavision anamorphic / 85mm / C-Series 等) — 通吃
- 底片/感光 (Kodak Vision3 500T / Cinestill 800T 等) — ✅ Flux/MJ;❌ Seedance
- 光比/燈光 (Rembrandt / 4:1 contrast / volumetric god rays 等) — 通吃
- 色彩分級 (teal-orange / bleach bypass / A24 indie 等) — 通吃
- 構圖/景別 (rule of thirds / medium close-up 等) — 通吃
- VFX/大氣 (halation / Tyndall effect / particles 等) — 通吃
- (Veo/Sora) 音訊三層 (Dialogue / SFX / Soundtrack)
### 禁用模式(通用 + 平台特定)
**通用原則:** **寧可 5 個具體 Discussion
Ask, share a tip or report that it does not work. Sorted by what people marked helpful.
No posts yet. Be the first to ask or share.
Posts here are written by users, not the PopNx team.
From GitHub
- GitHub stars
- 282
- Maker
- Hao0321
- License
- MIT
- Last code change
- Aug 9, 2026
- SKILL.md files in the repo (including translated copies and templates)
- 1
- Topics
- ai-image, ai-music, ai-video, claude-code, claude-skill, flux, generative-ai, kling, midjourney, prompt-engineering, seedance, sora