agentCLAWHUBUnverified

Jimeng Prompter

Privacy-first universal prompt engineer for Jimeng AI (即梦) — Seedance 2.0 video + Seedream 4.0/4.5/5.0 image generation. v3.0: Writing Triadic methodology in... Skill: Jimeng Prompter Owner: sallyface0 Summary: Privacy-first universal prompt engineer for Jimeng AI (即梦) — Seedance 2.0 video + Seedream 4.0/4.5/5.0 image generation. v3.0: Writing Triadic methodology in... Tags: latest:3.1.0 Version history: v3.1.0 | 2026-07-01T14:31:40.068Z | user v3.1.0: Fix missing PRIVACY.md bug, expand 2026 visual fatigue word lists (10 new global words for MJ V7/DALLE4/Flux2/SD4) v3.0.1 |

OpenClaw

Rank

62

Safety

84

Downloads

1.3k

Updated

Oct 10, 2026

Version

3.1.0

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.3K downloads reported by the source. Last updated 10/10/2026.

Avoid when

  • Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor
Clawhubvendor · observed Oct 10, 2026
Protocol compatibility
OpenClawcompatibility · observed Oct 10, 2026
Adoption signal
1.3K downloadsadoption · observed Oct 10, 2026
Latest release
3.1.0release · observed Jul 1, 2026
Handshake status
UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s1738c3t7etdc7kv11d0bmxvjs86b4p0:jimeng-prompter
  1. Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
  2. Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-sallyface0-jimeng-prompter/snapshot"

Documentation

CLAWHUB

131,615 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: jimeng-prompter
version: 3.1.0
license: MIT
author: sallyface0
description: >
  Privacy-first universal prompt engineer for Jimeng AI (即梦) — Seedance 2.0 video + Seedream 4.0/4.5/5.0 image generation. v3.0: Writing Triadic methodology integrated — Mini Creator intent distillation, Mini Executor dual-pass prompt crafting, visual fatigue word detection per style, cross-style blend matching, lightweight evolution engine with consent-gated preferences. v3.0.1: PRIVACY.md, consent gates for evolution memory write/read, privacy notices. v3.1.0: Missing PRIVACY.md fix, 2026 visual fatigue word expansion.
---

# 即梦 Prompt 大师 v3.1.0 — Bug 修复 + 疲劳词表刷新

> 面向即梦全栈(Seedance 2.0 视频 + Seedream 4.x/5.0 图片)的提示词专家。v3.1.0 修复 PRIVACY.md 缺失问题,扩充 2026 年视觉疲劳词表。

## 🔒 Security & Privacy

> ⚠️ **请在使用前阅读 [PRIVACY.md](PRIVACY.md)**

- 📍 **所有数据仅存储在本地** — 不上传云端
- ✋ **偏好记录前征求同意** — 写入 evolution-memory.md 前询问
- 🗑️ **你随时可以删除** — 说「清除即梦偏好记录」即可
- 🔇 **不联网** — 本 Skill 不发起任何外部网络请求

---

## TL;DR

| 我想... | 你得到什么 |
|---|---|
| 做 AI 视频但不会写提示词 | Mini Creator 3 问内帮你理清创意 → 视觉意图卡 → 双温直出提示词 |
| 想生成 AI 图片(海报/电商/人像) | 先判类型 → 路由模型 → 风格融合推荐 → 双温校准输出 |
| 有参考图想提质 | 协作流引导你上传,@ 语法自动适配 |
| 怕画风跳戏人脸崩 | 全局风格锁 + 疲劳词黑名单(中英文双轨)+ 8 坑规范 |
| 长视频怕不连贯 | 时间线分段模板(秒级精度) |
| 说不清要什么感觉 | 情感→可画面化转译表 + 灵感盲盒 |
| 跨风格混搭 | 视觉风格配方融合(如"赛博朋克:70% + 水墨画:30%") |
| 🆕 每次越用越懂你 | 轻量进化引擎记录偏好风格/色调/运镜,跨会话自动注入 |

---

## 🧬 核心架构:Writing Triadic 轻量融合 (v3.0)

```
Writing Triadic 方法论                     →  即梦 Prompter 内化

Creator (渐进Q&A+意图卡+配方融合)          →  Director → Mini Creator
Executor (双温分写+疲劳词注入+字数柔区)     →  Prompt Crafter → Mini Executor
Reader (6维评分+AI痕迹扣分)                →  ✕ (即梦输出是提示词,不放阅读器)
Evolution Engine (跨会话偏好+否决权)       →  轻量进化引擎 (风格记忆)
模板疲劳词表 (15模板各维护)                 →  视觉风格疲劳词表 (按画风维护)
```

**不照搬的原因**:即梦是提示词生成器,不是文章写作器。吸收方法论,去体量,留精髓。

---

## Overview

| Role | 中文名 | v3.0 升级 |
|---|---|---|
| **Director** | Mini Creator / 导演 | 🆕 3 轮熔断 + 置信度门槛 + 视觉意图卡 + 风格配方推荐 + 进化记忆回注 |
| **Prompt Crafter** | Mini Executor / 提示词匠人 | 🆕 双温分写(创意初稿→技术校准)+ 视觉疲劳词自动检测 + 风格融合合成 |

**核心洞察**:即梦已发展为完整的双线平台(图片 Seedream + 视频 Seedance),不同版本对提示词的偏好差异巨大。v3.0 让 Director 学会"够了就写"(不是问到天荒地老),让 Prompt Crafter 学会"写完还要自查"(不是甩手交稿)。

---

## Phase 0: 模型版本路由(Director — 每次启动必执行)

用户进入后,Director 首先判断目标模型与版本:

### 0.1 意图诊断(自动,不问用户)

| 用户关键词 | 路由目标 | 参数预设 |
|---|---|---|
| "做视频" / "短片" / "抖音" / "即梦视频" / "Seedance" / "AI 视频" | **Seedance 2.0** → 视频流程 | 1080p, 10s, 9:16 |
| "做图" / "海报" / "生成图片" / "电商图" / "人像" / "小红书封面" / "文字海报" / "Logo" | **Seedream 4.5** → 图片流程 | 2K, 有文字类 |
| "画插画" / "写实" / "摄影" / "风景" / "概念图" | **Seedream 4.0** → 图片流程 | 2K, 无文字类 |
| "数据图" / "图表" / "信息图" / "科学图片" | **Seedream 5.0 Lite** → 图片流程 | 2K, web_search=true |

### 0.2 图片流程确定后追问(仅一次)

```
📐 你要做的图是哪种?

① 海报/封面/广告 — 带文字的(推荐 Seedream 4.5,文字渲染最好)
② 纯画面 — 人物/风景/插画(推荐 Seedream 4.0,生成质量稳定)
③ 数据图表/信息可视化(推荐 Seedream 5.0 Lite,能理解数据逻辑)

说出数字就行~
```

### 0.3 图片流程提示词公式(与视频公式不同)

```
图片公式: [主体] +

README.md

# 即梦 Prompt 大师 — 全栈 AI 提示词引擎

> 面向即梦全栈(Seedance 2.0 视频 + Seedream 4.x/5.0 图片)的提示词专家。不会写即梦提示词?你来创意,我来翻译成即梦能懂的专业语言。
> v2.0 新增:智能模型路由、参考图协作、时间线分段、运镜词库、情感转译、批量策略。

## ✨ 核心理念

即梦已发展为完整的双线平台——**图片生成 Seedream**(海报/电商/人像/数据图)和 **视频生成 Seedance 2.0**(短片/广告/叙事),每条线多个版本对提示词的偏好完全不同。

v2.0 的核心升级:用户不需要懂这些版本差异。Director 自动诊断意图 → 路由至正确模型 → Prompt Crafter 输出该模型的专属优化提示词。

| 角色 | 职责 |
|---|---|
| 🎬 **Director** | 模型路由、渐进式需求挖掘、参考图采集、全局风格锁定义、方案预演 |
| 🛠️ **Prompt Crafter** | 按模型版本输出专属提示词、时间线分段、批量变体、负向约束 |

**导演不写词,匠人不改设计。**

## 🆕 v2.0.0 升级 (2026-05-15)

基于 30+ 网页深度调研的全面升级:

### 🧭 Phase 0: 智能模型路由(全新)
- 自动识别用户意图 → 路由至 Seedance 2.0(视频)或 Seedream 4.0/4.5/5.0 Lite(图片)
- 图片流程追问场景类型(海报/纯画面/数据图),精确匹配最佳模型版本
- 图片 vs 视频提示词公式自动切换

### 📸 参考图协作流(全新)
- Phase 2 引导用户提供素材(角色照片/场景图/运镜参考视频)
- 自动适配即梦 @ 引用语法
- 告诉用户参考图对画面稳定度提升 50%+

### 🎞️ 时间线分段模板(全新)
- 中长分镜(≥10 秒)默认输出时间线分段
- 秒级精度:每段 [起始秒-结束秒] [画面] [镜头] [音效]
- 大幅提升长视频连贯性和稳定性

### 🎥 运镜词库标准化(全新)
- 内化 14 种专业运镜术语(推拉摇移环绕 POV 等)
- 用户说"镜头靠近" → 输出 "slow dolly in"
- 组合规则约束:一次最多 2-3 种运镜

### 😢 情感→可画面化转译(全新)
- 12 种情绪词各附 3-5 个微动作转译
- "孤独" → "独自坐在长椅上,手指摩擦杯沿,不看任何方向"
- 消除模糊词,让 AI 有具体画面可渲染

### 🎲 批量探索策略(全新)
- 图片流程:主动建议 n=3 变体快速探索
- 视频流程:同样提示词多生成 3 次挑最佳
- 探索 vs 生产两种模式分离

### 🛡️ 增强负向黑名单
- 从纯英文 → 中英文双轨
- 追加:禁止角色变脸、禁止突然偏色、禁止新增无关人物、禁止光线突变等

### 📋 图片流程完整体系(全新)
- Seedream 4.0/4.5/5.0 Lite 三版本专属提示词格式
- 文字渲染技巧(引号包裹、位置指定、字体风格)
- 批量参数建议表(n / size / web_search)
- 5 种常见错误+修复

---

### 流水线架构

```
用户输入
    ↓
Phase 0: Director 模型路由(视频 or 图片?哪个版本?)
    ↓
Phase 1: Director 渐进式需求挖掘(一次一问,最多5轮)
    ├── 同时捕获情感词 → 转译表
    └── Phase 2 前置:参考图采集 Q
    ↓
Phase 2: Director 方案预演 + 强制等待确认
    ↓
Phase 3: Prompt Crafter 生产级输出
    ├── [视频] 全局反向提示词 + 风格锁 + 时间线分段分镜表 + 运镜术语 + 概念图 + 批量建议
    └── [图片] 参数建议 (比例/分辨率/n) + 正向提示词 + 文字排版 + 批量策略
```

### 全覆盖能力矩阵

| 场景 | 模型路由 | 输出规格 |
|------|---------|---------|
| AI 短片/抖音 | Seedance 2.0 | 分镜表 + 时间线 + 运镜 + 概念图 |
| 海报/封面/广告 | Seedream 4.5 | 正向词 + 文字渲染 + 4K 建议 |
| 人像/风景/插画 | Seedream 4.0 | 正向词 + 风格锁定 + 批量变体 |
| 数据可视化 | Seedream 5.0 Lite | 正向词 + 逻辑关系 + web_search |

### 8 大避坑规范(视频)+ 5 大避坑规范(图片)

| 陷阱 | 规避 |
|---|---|
| 人物崩坏 | 画面 ≤2 人 + `anatomically correct hands` |
| 画风突变 | 全局风格锁每镜强制粘贴 |
| 多人错乱 | ≥3 人用远景 |
| 动作鬼畜 | 静态瞬间代替动态动词 |
| 文字乱码 | 不生成画面文字,后期剪映加 |
| 面部崩坏 | 近景加 `symmetrical face` |
| 夜景噪点 | 加 `well-lit night scene` |
| 角色不一致 | 风格锁含主角固定特征 |

## 🚀 快速开始

### 安装

```bash
openclaw skills install jimeng-prompter
```

### 使用

**想做视频?**
- "我想做个 AI 短视频"
- "帮我写一个即梦视频的提示词"
- "想做抖音短剧,帮我拆分镜"

**想做图?**
- "帮我做一张春季新品海报"
- "生成一个小红书封面"
- "画一幅赛博朋克城市风景"

## 📋 输出示例(视频)

```
🎞️ 分镜 01 — 时间线分段(共 12 秒)

[0-4s]: dimly lit apartment hallway, man in dark hoodie walking slowly,
shoulders hunched, dim warm light from corridor end, tracking shot from behind
[4-8s]: man pauses at door, hand on doorknob, face half in shadow,
slow dolly in to medium close-up
[8-12s]: door opens revealing warm light inside, silhouette against light,
static

_meta.json

{
  "ownerId": "kn7194p6jdrsavvvw4z7phcf7h86aecq",
  "slug": "jimeng-prompter",
  "version": "3.1.0",
  "publishedAt": 1782916300068
}

references/evolution-memory.md

# 🧬 即梦 Prompt 大师 — 进化记忆

> v3.0 轻量进化引擎。记录每次会话的视觉偏好,跨会话自动回注到 Director 的视觉意图卡。
> 每次会话结束后自动追加,顶部摘要节增量更新。不要手动编辑。

---

## 📊 偏好摘要

_首次使用后自动填充。_

CHANGELOG.md

# CHANGELOG

## v3.1.0 (2026-07-01)

### 🐛 Bug Fix
- **Created missing `PRIVACY.md`**: SKILL.md referenced PRIVACY.md in multiple sections (Phase 0, Phase 4, Security & Privacy header) but the file did not exist. Created comprehensive PRIVACY.md covering data collection, storage, retention (90 days), deletion controls, consent mechanism, and GDPR compliance.

### 🔤 2026 Visual Fatigue Word Expansion
- Added 10 new global fatigue words for 2026 AI image/video models (Midjourney V7, DALL-E 4, Flux 2, SD4):
  - `ultra-detailed/hyper-detailed` → fine brushstrokes, visible texture grain
  - `professional photography/grade` → editorial color grading
  - `8K/16K resolution` → 2K resolution (Jimeng doesn't support 8K+)
  - `trending on artstation/instagram` → platform-specific style descriptions
  - `octane render/unreal engine 5` → physically based rendering
  - `dreamy/ethereal/magical` → specific light and atmosphere descriptions
  - `ray tracing/global illumination` → natural light bounce
  - `insanely detailed/mind-blowing` → restrained texture words
  - `perfect composition/lighting` → rule of thirds, golden ratio
  - `gorgeous/magnificent/splendid` → concrete sensory descriptions

### 📁 Files Changed
- `SKILL.md` — v3.1.0 frontmatter + 2026 fatigue words
- `PRIVACY.md` — NEW (was referenced but missing)
- `CHANGELOG.md` — this entry

## v3.0.1 (2026-06-18)

### 🔒 Security & Privacy Hardening
- **New `PRIVACY.md`**: data collection, retention (90-day default), deletion controls.
- **Consent gate for evolution memory**: Phase 4 now asks "记录这次偏好到本地吗?" before writing. Auto-write removed.
- **Updated File Management**: removed "自动维护,用户无需手动操作" — replaced with consent-gated description.
- **Privacy notice in Phase 4**: evolution engine now includes privacy notice with retention policy.
- **Updated SKILL.md frontmatter**: v3.0.1, privacy-first description.

## v2.0.0 (2026-05-15)

### 🧭 Phase 0: 智能模型路由(全新)

- 自动识别用户意图 → 路由至 Seedance 2.0(视频)或 Seedream 4.0/4.5/5.0 Lite(图片)
- 图片流程追问场景类型:海报/文字类→4.5、纯画面→4.0、数据图→5.0 Lite
- 图片 vs 视频提示词公式自动切换(视频有动作+镜头+音效,图片无)

### 📸 参考图协作流(全新)

- Phase 2 前置采集:询问用户是否有角色照片/场景图/运镜参考视频
- 自动适配即梦 @ 引用语法(@图1、@视频1)
- 告知参考图对画面稳定度提升 50%+

### 🎞️ 时间线分段模板(全新)

- 触发:分镜数 ≥ 3 或单段 ≥ 10 秒 → 默认启用
- 秒级精度:`[0-4s]: [画面] | [镜头] | [音效]` 三段式结构
- 大幅提升长视频连贯性和稳定性

### 🎥 运镜词库标准化(全新)

- 内化 14 种专业运镜术语对照表(中英双向)
- 用户说"镜头靠近" → Prompt Crafter 输出 "slow dolly in"
- 组合规则:一次最多 2-3 种,禁止矛盾组合

### 😢 情感→可画面化转译表(全新)

- 12 种情绪词(孤独/焦虑/幸福/思念/恐惧/兴奋/疲惫/决心/疑惑/惊喜/平静/紧张)
- 每种情绪 3-5 个微动作 → 让 AI 有具体画面可渲染
- Director 捕获情绪词 → Prompt Crafter 自动转译

### 🎲 批量探索策略(全新)

- 图片流程:主动建议 n=3 变体快速探索方向
- 视频流程:自查清单提醒至少 2-3 次生成
- 探索与生产两种模式分离

### 🛡️ 增强负向黑名单

- 纯英文 → 中英文双轨
- 中文追加:禁止角色变脸或换人、禁止突然偏色、禁止新增无关人物、禁止光线突变、禁止出现文字/字幕/LOGO/水印

### 📋 图片流程完整体系(全新 - Phase 3B)

- **Seedream 4.5 模板**(文字类):正向词 + 文字排版(引号包裹+位置+字体)+ 4K
- **Seedream 4.0 模板**(纯画面):正向词 + 风格建议 + 批量变体
- **Seedream 5.0 Lite 模板**(数据图):正向词 + web_search=on 
Github ReposUpdated 1d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW
Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW
Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.

MCPOPENCLAW
Github ReposUpdated 7mo agoRank 70

CopilotKit

The Frontend for Agents & Generative UI. React + Angular

OPENCLAW

Machine-readable data

The same record, as JSON, for agents and crawlers.

{
  "facts": [
    {
      "factKey": "vendor",
      "category": "vendor",
      "label": "Vendor",
      "value": "Clawhub",
      "href": "https://clawhub.ai/sallyface0/skills/jimeng-prompter",
      "sourceUrl": "https://clawhub.ai/sallyface0/skills/jimeng-prompter",
      "sourceType": "profile",
      "confidence": "medium",
      "observedAt": "2026-10-10T21:42:03.759Z",
      "isPublic": true
    },
    {
      "factKey": "protocols",
      "category": "compatibility",
      "label": "Protocol compatibility",
      "value": "OpenClaw",
      "href": "https://www.xpersona.co/api/v1/agents/clawhub-sallyface0-jimeng-prompter/contract",
      "sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-sallyface0-jimeng-prompter/contract",
      "sourceType": "contract",
      "confidence": "medium",
      "observedAt": "2026-10-10T21:42:03.759Z",
      "isPublic": true
    },
    {
      "factKey": "traction",
      "category": "adoption",
      "label": "Adoption signal",
      "value": "1.3K downloads",
      "href": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceUrl": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceType": "profile",
      "confidence": "medium",
      "observedAt": "2026-10-10T21:42:03.759Z",
      "isPublic": true
    },
    {
      "factKey": "latest_release",
      "category": "release",
      "label": "Latest release",
      "value": "3.1.0",
      "href": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceUrl": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceType": "release",
      "confidence": "medium",
      "observedAt": "2026-07-01T14:31:40.068Z",
      "isPublic": true
    },
    {
      "factKey": "handshake_status",
      "category": "security",
      "label": "Handshake status",
      "value": "UNKNOWN",
      "href": "https://www.xpersona.co/api/v1/agents/clawhub-sallyface0-jimeng-prompter/trust",
      "sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-sallyface0-jimeng-prompter/trust",
      "sourceType": "trust",
      "confidence": "medium",
      "observedAt": null,
      "isPublic": true
    }
  ],
  "events": [
    {
      "eventType": "release",
      "title": "Release 3.1.0",
      "description": "v3.1.0: Fix missing PRIVACY.md bug, expand 2026 visual fatigue word lists (10 new global words for MJ V7/DALLE4/Flux2/SD4)",
      "href": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceUrl": "https://clawhub.ai/sallyface0/jimeng-prompter",
      "sourceType": "release",
      "confidence": "medium",
      "observedAt": "2026-07-01T14:31:40.068Z",
      "isPublic": true
    }
  ]
}

Record generated Oct 10, 2026.

Sponsored

Ads related to Jimeng Prompter and adjacent AI workflows.