← 返回 Skills 市场
246
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install tianshu-image
功能描述
使用阿里云通义万相 (DashScope) 生成图片。Use when: 用户需要根据文字描述生成图片；用户说「画一张」「生成图片」「文生图」。 NOT for: 图片编辑、风格迁移、其他非文生图场景。
使用说明 (SKILL.md)
天树文生图 (tianshu-image)
使用阿里云通义万相 API 根据文字描述生成图片，Node.js 实现，无需 Python/uv。
When to Run
- 用户说「画一张」「生成图片」「文生图」「根据这段描述生成图」
- 用户提供图片描述，需要 AI 生成对应图片
前置配置
在 ~/.openclaw/openclaw.json 的 skills.entries.tianshu-image 中配置，或设置环境变量：
DASHSCOPE_API_KEY- 阿里云 DashScope API Key（从 https://dashscope.console.aliyun.com/ 获取）- 或
models.providers.bailian.apiKey（若已配置百炼）
Workflow
- 确认用户需要生成图片
- 执行脚本：
node ~/.openclaw/skills/tianshu-image/scripts/generate_image.js --prompt "描述" [--size 1664*928] - 脚本输出
MEDIA_URL: \x3Curl>或MEDIA: \x3C本地路径> - 用 Markdown 展示图片：
参数
--prompt/-p：图片描述（必填）--size/-s：尺寸，默认1664*928。可选：1024*1024、720*1280、1280*720--model/-m：模型，默认qwen-image-max。可选：qwen-image-turbo、qwen-image-plus-2026-01-09--filename/-f：保存到本地文件路径（可选）--negative-prompt：负向提示词--no-prompt-extend：禁用自动提示词增强--watermark：添加水印
Output
- 未指定
--filename时输出MEDIA_URL: \x3Curl>，直接展示 URL - 指定
--filename时输出MEDIA: \x3C本地路径>，表示已保存到本地
安全使用建议
This skill will send the text prompt (and negative prompt) to Alibaba DashScope and uses the provided DASHSCOPE_API_KEY to authenticate. Only install if you trust that service and the API key you provide. Note the SKILL.md mentions alternate config locations that the script does not actually read—verify you set DASHSCOPE_API_KEY or pass --api-key when running. Be aware generated images may be returned as external URLs (the script prints MEDIA_URL) — if you prefer not to expose images to third-party hosting, use --filename to save locally. Finally, restrict the API key scope where possible and avoid supplying highly sensitive prompts or data to the skill.
功能分析
Type: OpenClaw Skill
Name: tianshu-image
Version: 1.0.0
The skill provides a legitimate interface for generating images using Alibaba's DashScope (通义万相) API. The core logic in `scripts/generate_image.js` handles API requests, argument parsing, and optional file saving as described in the documentation. No evidence of malicious intent, data exfiltration, or prompt injection was found; the script's behavior is entirely consistent with its stated purpose.
能力评估
Purpose & Capability
Name and description match the included script: the code sends a text prompt to Alibaba DashScope and returns or saves the generated image. Minor mismatch: SKILL.md mentions an alternative config key (models.providers.bailian.apiKey) and suggests storing config in ~/.openclaw/openclaw.json, but the script only reads DASHSCOPE_API_KEY or a passed --api-key; this is a documentation inconsistency, not a functional mismatch with purpose.
Instruction Scope
SKILL.md instructs running the included Node.js script and describes expected outputs — which align with the code. It also references reading openclaw.json or a different provider key, but the script does not read those paths; otherwise the instructions do not ask the agent to read unrelated files or exfiltrate arbitrary data.
Install Mechanism
No install spec and no external downloads; the skill is instruction+script only (package.json has no dependencies). This is low-risk from an installation perspective.
Credentials
Only DASHSCOPE_API_KEY (or --api-key) is required, which is proportional to a skill that calls an external image-generation API. No unrelated credentials or broad system secrets are requested.
Persistence & Privilege
Skill does not request persistent/global privileges (always is false). It can write image files only when the user supplies a --filename; it does not modify other skills or system-wide configuration.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install tianshu-image - 安装完成后，直接呼叫该 Skill 的名称或使用
/tianshu-image触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Node.js 实现，替代 qwen-image，无需 Python/uv
元数据
常见问题
Tianshu Image 是什么？
使用阿里云通义万相 (DashScope) 生成图片。Use when: 用户需要根据文字描述生成图片；用户说「画一张」「生成图片」「文生图」。 NOT for: 图片编辑、风格迁移、其他非文生图场景。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 246 次。
如何安装 Tianshu Image？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install tianshu-image」即可一键安装，无需额外配置。
Tianshu Image 是免费的吗？
是的，Tianshu Image 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Tianshu Image 支持哪些平台？
Tianshu Image 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Tianshu Image？
由 tianshu（@wangshengli0421）开发并维护，当前版本 v1.0.0。
推荐 Skills