即梦AI视频生成

即梦AI视频生成工具（带声音版本），通过火山引擎API自动生成带音频的高质量视频。支持文生视频、图生视频，适用于短视频内容创作。

Security Scan

The skill's purpose (Volcengine/Dreamina video+audio generation) matches the instructions and requested credentials, but there are packaging and documentation inconsistencies that warrant caution before installing or granting keys.

Purpose & Capability

SKILL.md and skill.yaml consistently describe a Volcengine (火山引擎) Dreamina video-generation integration (models, API endpoints, task flow). Requested capabilities (generate/query tasks, model choice) align with the stated purpose.

Instruction Scope

Instructions are primarily curl examples calling ark.cn-beijing.volces.com and describe storing credentials under ~/.openclaw/.credentials/volcengine-dreamina.env. No unrelated file reads or broad data-collection steps are present. Minor inconsistency: curl examples use Authorization: Bearer ${API_KEY} while configuration/skill.yaml specify VOLCENGINE_ACCESS_KEY_ID and VOLCENGINE_SECRET_ACCESS_KEY—unclear how those map to the Authorization header.

Install Mechanism

This is an instruction-only skill with no install spec and no code files (lowest install risk). Note: README claims "已自动安装，无需额外操作" and references a CLI binary path (~/.openclaw/skills/jimeng-video/jimeng-video), but no installer or binary is provided—documentation/packaging mismatch.

Credentials

The only sensitive items described are Volcengine API credentials (VOLCENGINE_ACCESS_KEY_ID and VOLCENGINE_SECRET_ACCESS_KEY) which are proportionate to the skill's functionality. However, registry metadata at the top listed 'required env vars: none' while skill.yaml requires two secret variables—this metadata inconsistency should be resolved before trusting the package.

Persistence & Privilege

The skill does not request always:true or other elevated persistent privileges. It suggests storing credentials in a skill-specific path (~/.openclaw/.credentials/...), which is normal for credential use and does not indicate cross-skill configuration changes.

What to consider before installing

This skill appears to do what it claims (call Volcengine/Dreamina to generate videos with audio), but there are a few red flags you should verify before installing or providing credentials: - Packaging/documentation mismatches: the registry metadata claims no env vars while skill.yaml requires VOLCENGINE_ACCESS_KEY_ID and VOLCENGINE_SECRET_ACCESS_KEY; README says the binary is "already installed" but no install spec or binary is included. Ask the publisher to clarify how the skill is delivered and invoked. - Auth variable inconsistency: curl examples use Authorization: Bearer ${API_KEY} but the declared secrets are an access key ID and secret. Confirm how to derive the actual Authorization header and whether the skill needs a single API token or the two keys. - Credential scope: create and use a dedicated Volcengine API key with the minimum permissions (only the Dreamina/generation service) and make it revocable. Do not reuse high-privilege or long-lived account keys. - Network/domain sanity check: the examples call ark.cn-beijing.volces.com (Volcengine-like domain). Confirm this is the legitimate endpoint for your account/region before sending keys or content. - Operational caution: because this is instruction-only (no bundled code), risk of arbitrary code on disk is low, but the skill will cause network requests that could send prompt/content and your supplied media to the vendor. If you need to keep prompts or media private, review the vendor's data usage policy or test with non-sensitive content. If the publisher can resolve the env-var and install inconsistencies and you can limit the API key scope, the skill is reasonable to use. If they cannot explain these mismatches, treat the package as untrusted.

jimeng-video Skill v2.0

即梦AI（Dreamina）视频生成工具 - 带声音版本

🎉 v2.0 重大更新

✅ 支持生成带声音的视频！

默认使用 doubao-seedance-1-5-pro-251215 模型，生成的视频自带音频（环境音/背景音乐/音效），无需后期配音！

---

功能特性

• 🎵 文生音视频：输入文字描述，自动生成带声音的视频
• 🎵 图生音视频：上传图片，生成带声音的动态视频
• 🎨 多模型支持：支持多个视频生成模型（带声音/纯视频）
• ⚡ 批量生成：支持一次性生成多个视频
• 📊 任务管理：创建任务、查询状态、获取结果

---

支持的模型

🌟 推荐模型（默认）

模型ID	类型	特点	音频
doubao-seedance-1-5-pro-251215	文生音视频	生成视频自带声音，效果最好	✅ 带声音

其他可用模型

模型ID	类型	特点	音频
doubao-seedance-1-0-pro-250528	视频生成	Pro版，质量高	❌ 无声
doubao-seedance-1-0-lite-t2v-250428	文生视频	轻量版，速度快	❌ 无声
doubao-seedance-2-0-260128	视频生成	最新版，功能强	❌ 无声

---

快速开始

1. 生成带声音的视频（默认）

# 使用默认模型（带声音）
jimeng-video generate \
  -p "一匹马高高跃起，四个字从左到右书写，水墨风格，文字内容是马年大吉" \
  -o output.mp4

# 生成的视频将包含：画面 + 自动生成的音效/音乐

2. 指定纯视频模型（无声）

# 使用旧版模型（无声，如需纯画面）
jimeng-video generate \
  -m doubao-seedance-1-0-lite-t2v-250428 \
  -p "夕阳下的海滩，金色的阳光洒在波浪上" \
  -o beach.mp4

3. 批量生成

# 从文件批量生成
jimeng-video batch \
  --file prompts.txt \
  --output-dir ./videos/

4. 查询任务状态

# 查询指定任务
jimeng-video status --task-id cgt-20260228005550-95t42

---

API 调用示例

生成带声音的视频

curl -X POST "https://ark.cn-beijing.volces.com/api/v3/contents/generations/tasks" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ${API_KEY}" \
  -d '{
    "model": "doubao-seedance-1-5-pro-251215",
    "content": [
      {
        "type": "text",
        "text": "你的提示词内容"
      }
    ]
  }'

查询结果

curl -X GET "https://ark.cn-beijing.volces.com/api/v3/contents/generations/tasks/{task_id}" \
  -H "Authorization: Bearer ${API_KEY}"

---

提示词技巧

带声音视频的有效提示词结构

[场景描述] + [动作/事件] + [氛围/情绪] + [声音提示]

示例（带声音暗示）：

• ✅ "热闹的台球厅，球杆击球声，人群的欢呼，专业比赛氛围"
• ✅ "海浪拍打沙滩，海鸥叫声，轻松惬意的夏日午后"
• ✅ "繁忙的城市街道，车声喇叭声，霓虹灯闪烁的夜晚"

声音相关关键词（帮助模型生成更好音效）：

• 环境音："嘈杂的人群声"、"雨声"、"风声"
• 动作音："击球声"、"键盘敲击声"、"脚步声"
• 情绪音："欢快的音乐"、"紧张的氛围"、"舒缓的旋律"

---

配置说明

环境变量

在 ~/.openclaw/.credentials/volcengine-dreamina.env 中配置：

VOLCENGINE_ACCESS_KEY_ID=your-access-key-id
VOLCENGINE_SECRET_ACCESS_KEY=your-secret-key

获取API Key

1. 登录火山引擎控制台：console.volcengine.com
2. 进入「访问控制」→「API Key管理」
3. 创建新的API Key
4. 开通即梦AI视频生成服务

---

模型对比

维度	1.5 Pro（带声音）	1.0 Lite（无声）
音频	✅ 自动生成	❌ 需后期添加
画质	1080p	720p/1080p
生成速度	30-40秒	15-25秒
适用场景	短视频成品	素材/需后期配音
价格	较高	较低

建议：

• 成品视频 → 使用 1.5 Pro（带声音）
• 素材剪辑 → 使用 1.0 Lite（无声）

---

返回结果

成功响应（带声音模型）

{
  "id": "cgt-20260228140147-6t2sz",
  "model": "doubao-seedance-1-5-pro-251215",
  "status": "succeeded",
  "content": {
    "video_url": "https://..."
  },
  "generate_audio": true,
  "resolution": "720p",
  "duration": 5,
  "framespersecond": 24
}

关键字段：

• generate_audio: true - 表示包含音频生成
• video_url - 视频下载链接（包含音视频）
• duration - 视频时长（秒）

---

使用场景

• 短视频创作：直接生成带声音的成品视频
• 广告制作：快速产出带音效的广告素材
• 内容运营：自动生成带背景音的短视频
• 创意实验：AI自动生成完整音视频内容

---

注意事项

1. 音频质量：自动生成的音频是AI合成的环境音/音乐，不是真实录音
2. 生成时间：带声音版本比无声版本慢约50%（30-40秒 vs 15-25秒）
3. 定制音频：如需特定音乐，建议用无声版本+后期添加
4. 音量平衡：自动音频音量适中，如需调整请后期处理

---

故障排查

模型未开通

错误： ModelNotOpen 解决： 在火山引擎控制台开通即梦AI 1.5 Pro服务

生成超时

现象： 任务长时间 running 解决： 带声音模型生成较慢，请耐心等待（最长2分钟）

音频不满意

解决：

• 调整提示词中的声音描述
• 或使用无声模型+后期添加自定义音乐

---

更新日志

v2.0.0 (2026-02-28)

• ✅ 升级默认模型为 doubao-seedance-1-5-pro-251215
• ✅ 支持生成带声音的视频
• ✅ 优化提示词文档，增加声音相关技巧
• ✅ 添加模型对比说明

v1.0.0 (2026-02-27)

• 初始版本
• 支持基础视频生成功能

---

让视频创作更简单，一步到位生成带声音的成品！ 🎬🎵