← Back to Skills Marketplace
felixhsp

jimeng generator

by FelixHsp · GitHub ↗ · v0.1.1
cross-platform ⚠ suspicious
434
Downloads
0
Stars
7
Active Installs
2
Versions
Install in OpenClaw
/install jimeng-generator
Description
即梦 4.0 图片生成器,通过文本描述生成高质量图片,支持多图编辑与智能比例。
README (SKILL.md)

即梦 4.0 图片生成器

基于火山引擎即梦 AI 4.0 生成图片。一条命令完成:提交任务 → 等待完成 → 保存图片。

配置凭证

使用前需要配置火山引擎凭证。请按以下步骤操作:

  1. 前往 火山引擎控制台 → 访问控制 → 密钥管理,获取 Access Key 和 Secret Key
  2. 在项目根目录创建 .env 文件,写入你的凭证:
VOLCENGINE_AK=你的AccessKey
VOLCENGINE_SK=你的SecretKey

如果使用 STS 临时凭证,改为填写 AK + Token:

VOLCENGINE_AK=你的AccessKey
VOLCENGINE_TOKEN=你的SecurityToken

项目已提供 .env.example 模板,也可以直接复制后修改:

cp .env.example .env

.env 文件已被 .gitignore 忽略,不会提交到仓库,请放心填写真实凭证。

安装

npm install

基本用法

npx ts-node scripts/generate.ts "提示词"

脚本会自动提交 → 轮询 → 保存图片到 ./output/

完整用法

npx ts-node scripts/generate.ts "提示词" [选项]

选项

选项 说明 默认值
--images \x3Curl,...> 参考图片 URL,逗号分隔,最多 10 张
--width \x3Cn> 输出宽度 自动
--height \x3Cn> 输出高度 自动
--size \x3Cn> 输出面积 自动
--scale \x3C0-1> 文本影响程度 0.5
--single 强制单图输出 false
--out \x3Cdir> 输出目录 ./output
--no-save 不保存,只输出 URL false
--interval \x3Cms> 轮询间隔 3000
--timeout \x3Cms> 最大等待时间 180000
--debug 调试模式 false

使用示例

文生图

npx ts-node scripts/generate.ts "水墨山水画"

指定尺寸

npx ts-node scripts/generate.ts "赛博朋克城市" --width 2560 --height 1440

图片编辑

npx ts-node scripts/generate.ts "背景换成星空" --images "https://example.com/photo.jpg"

多图组合

npx ts-node scripts/generate.ts "合成一张合照" --images "https://a.jpg,https://b.jpg"

强制单图 + 高影响

npx ts-node scripts/generate.ts "精细插画风格的城堡" --single --scale 0.8

输出格式

脚本结果通过 stdout 输出一行 JSON,便于解析。请用其中的 files 数组直接作为“生成结果”返回给用户(本地文件路径,可用来展示或附件),不要只贴直链或把本地路径藏在文案里。

成功时示例:

{
  "success": true,
  "taskId": "7392616336519610409",
  "prompt": "水墨山水画",
  "count": 1,
  "files": ["/path/to/output/1.png"],
  "urls": ["https://..."]
}
字段 说明
files 本地文件路径数组,即生成图片的保存位置,应作为主结果直接返回给用户
urls 图片直链(可选),仅作备用
taskId 任务 ID
count 图片张数

集成建议:解析 stdout 该行 JSON,若 success === true,则把 files 中的路径作为“已生成的图片”直接提供给用户(例如展示图片或作为附件),无需再组织成“直链在这:… 本地文件在:…”的文案。

失败时错误信息在 stderr,JSON 形如:

{
  "success": false,
  "error": { "code": "FAILED", "message": "错误描述" }
}

即梦 4.0 特性

  • 智能比例:可在 prompt 中描述比例,模型自动适配最优宽高
  • 多图输入:最多 10 张参考图,支持图片编辑和多图组合
  • 多图输出:单次最多输出 15 张关联图片
  • 4K 输出:支持从 1K 到 4K 分辨率
  • 中文增强:显著提升中文生成准确率

推荐尺寸

分辨率 1:1 4:3 3:2 16:9 21:9
2K 2048×2048 2304×1728 2496×1664 2560×1440 3024×1296
4K 4096×4096 4694×3520 4992×3328 5404×3040 6198×2656
Usage Guidance
This package appears to be a legitimate VolcEngine (Jimeng) image generator: it will ask you to supply VOLCENGINE_AK and VOLCENGINE_SK (or a temporary token) in a .env file and then run a TypeScript script that signs requests and calls visual.volcengineapi.com, saving images to ./output. Before installing or running it: 1) note the registry metadata did NOT declare required env vars — treat that as a packaging oversight and verify you are intentionally providing VolcEngine credentials; 2) prefer using a short-lived STS token or least-privilege key and rotate/revoke keys after testing; 3) run npm install and execute the script in an isolated environment (container or sandbox) so network calls and file writes are contained; 4) inspect scripts/generate.ts yourself (or a reviewer) — it is the real runtime code and currently does not contain obfuscated or unexpected endpoints; 5) verify dependency sources if you are concerned about supply-chain risk (npm mirror entries are present in package-lock.json). If any of these checks make you uncomfortable, do not provide long-lived credentials or run the code on sensitive hosts.
Capability Analysis
Type: OpenClaw Skill Name: jimeng-generator Version: 0.1.1 The skill bundle is a legitimate integration for the VolcEngine Jimeng AI 4.0 image generation service. The core logic in `scripts/generate.ts` correctly implements the VolcEngine HMAC-SHA256 request signing protocol and handles asynchronous task submission, polling, and image downloading/saving. The `SKILL.md` provides clear, non-malicious instructions for the AI agent to execute the generator and return local file paths to the user. No evidence of data exfiltration, unauthorized network calls, or malicious prompt injection was found.
Capability Assessment
Purpose & Capability
The name/description, README, SKILL.md and scripts/generate.ts consistently implement a VolcEngine (jimeng) image generator that requires VOLCENGINE_AK/VOLCENGINE_SK (or VOLCENGINE_TOKEN). However the registry metadata lists no required environment variables or primary credential — that is a clear mismatch between declared requirements and actual code/instructions.
Instruction Scope
SKILL.md instructs the agent/user to create a .env with VolcEngine credentials, run npm install, and invoke the TypeScript script which submits tasks, polls status, and writes images to ./output. The instructions and the script operate within the stated purpose (submit tasks to visual.volcengineapi.com and save results). The script reads only the project .env and writes output images; it does not try to read unrelated system paths or exfiltrate data to unexpected endpoints.
Install Mechanism
No install spec in registry (instruction-only), but the package includes a Node.js script and package.json that requires running npm install (axios, ts-node, typescript). Installing npm deps is normal for this tool; dependencies come from standard npm registries/mirrors (package-lock uses a mirror). There is no remote arbitrary download or extract of unknown archives in the skill files. Still, because code will be executed locally, users should install and run it in a controlled environment.
Credentials
The code and SKILL.md legitimately require VolcEngine credentials (VOLCENGINE_AK, VOLCENGINE_SK, optionally VOLCENGINE_TOKEN). That access is proportional to the stated purpose. The concern is that the skill metadata did not declare these required env vars, so an installer/agent might not realize credentials are needed — increasing the chance a user will supply secrets without noticing. No other unrelated credentials are requested.
Persistence & Privilege
The skill does not request always:true and does not modify other skills or system-wide settings. It runs as a local CLI script; autonomy (agent-initiated invocation) is enabled by default but not combined with other red flags.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install jimeng-generator
  3. After installation, invoke the skill by name or use /jimeng-generator
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.1
更新技能文档与使用说明
v0.1.0
- 初始发布:即梦 4.0 图片生成器,支持通过文本描述生成高质量图片。 - 支持多图编辑、智能比例及多种分辨率(1K–4K),显著提升中文生成效果。 - 提供命令行脚本一键完成任务提交、自动轮询与本地图片保存。 - 支持传入最大 10 张参考图,单次最多生成 15 张图片,兼容多图合成与编辑。 - 输出结果为结构化 JSON,本地路径自动提取,便于集成与展示。 - 全面文档说明,覆盖凭证配置、常用命令与参数说明。
Metadata
Slug jimeng-generator
Version 0.1.1
License
All-time Installs 7
Active Installs 7
Total Versions 2
Frequently Asked Questions

What is jimeng generator?

即梦 4.0 图片生成器,通过文本描述生成高质量图片,支持多图编辑与智能比例。 It is an AI Agent Skill for Claude Code / OpenClaw, with 434 downloads so far.

How do I install jimeng generator?

Run "/install jimeng-generator" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is jimeng generator free?

Yes, jimeng generator is completely free (open-source). You can download, install and use it at no cost.

Which platforms does jimeng generator support?

jimeng generator is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created jimeng generator?

It is built and maintained by FelixHsp (@felixhsp); the current version is v0.1.1.

💬 Comments