← 返回 Skills 市场
72
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install nano-banana-skills
功能描述
Generate images using Google Gemini models (Nano Banana 2 / gemini-3-pro-image-preview). Use when the user asks to create, generate, or make an image, pictur...
安全使用建议
This skill appears to implement the advertised image-generation functionality, but confirm the following before installing or running it:
- An API key is required at runtime: the scripts expect WISGATE_KEY (SKILL.md/README mention this), but the registry metadata does not declare any required environment variables. Ask the publisher to correct the registry metadata or declare the env var so you know a secret is needed.
- There is a minor typo/inconsistency in the SKILL.md (references to WISDOM_GATE_KEY in one place). Make sure you set the correct variable (WISGATE_KEY) and avoid accidentally pasting your key into the wrong place.
- The scripts call https://api.wisgate.ai and will send input images and prompts to that external service; only provide images and prompts you are comfortable being transmitted to that API. Review Wisgate's privacy/security policy if you will upload sensitive images.
- The skill saves conversation history to conversation.json in the working directory and writes generated images to disk. If this is sensitive data, run the scripts in a controlled/sandbox directory or delete the history file after use.
- Because the registry metadata and SKILL.md disagree about required credentials, request clarification from the owner (or inspect the code yourself, which you can: both scripts are included) before granting any API key.
If you trust the Wisdom Gate service and are comfortable storing an API key as an env var, the skill is functionally coherent; the main issue is the metadata/documentation mismatch which should be fixed.
功能分析
Type: OpenClaw Skill
Name: nano-banana-skills
Version: 1.0.0
The skill facilitates image generation and refinement by interfacing with a third-party API (api.wisgate.ai). It reads the 'WISGATE_KEY' environment variable and transmits it to this external endpoint, while also performing local file read/write operations for images and conversation history (scripts/generate_image.py, scripts/refine_image.py). While these actions are aligned with the stated purpose, the use of a non-official API claiming access to non-existent or future-dated models (e.g., 'gemini-3.1-flash-image-preview') and the handling of sensitive API keys constitute risky capabilities without clear evidence of intentional malice.
能力评估
Purpose & Capability
Name, description, README, SKILL.md and the two scripts all implement an image-generation workflow against the Wisdom Gate (wisgate.ai) API (text→image and image→image). There are no unrelated API keys or unrelated binaries requested; the required functionality (upload input images, call external image-generation API, save outputs and conversation history) is consistent with the stated purpose.
Instruction Scope
Runtime instructions and scripts instruct the agent to call https://api.wisgate.ai and to read input image files and write outputs and a conversation.json history file. This scope is expected for a multi-turn image generator, but the SKILL.md contains a truncated/typo'd workflow line referencing a different env var name (WISDOM_GATE_KEY) which could cause confusion. The scripts only access local input/output files and the conversation history; they do not attempt to read unrelated system paths.
Install Mechanism
No install spec is provided (instruction-only install), and included Python scripts require only the requests library. There is no remote download of arbitrary archives or unusual install steps. Risk from installation is low, but the code will run network calls when invoked.
Credentials
The scripts and SKILL.md require a Wisdom Gate API key (environment variable referenced as WISGATE_KEY), which is proportionate to the purpose. However, the registry metadata claims no required env vars, and the SKILL.md and README contain inconsistent/typo'd references (WISGATE_KEY vs WISDOM_GATE_KEY). The mismatch between declared metadata and actual code means an API key will be needed at runtime even though the registry advertises none—this is an incoherence that could lead to accidental key exposure or confusion.
Persistence & Privilege
The skill does not request permanent platform presence (always: false) and does not modify other skills or system-wide agent settings. It writes conversation history (conversation.json) and image output files to the working directory, which is expected for multi-turn refinement and not an elevated privilege request.
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 nano-banana-skills - 安装完成后，直接呼叫该 Skill 的名称或使用
/nano-banana-skills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
Nano Banana Gemini Image Generation Skill v1.0.0
- Introduces image generation using Google Gemini models via the Wisdom Gate API.
- Supports text-to-image and image-to-image generation (including multi-image input and modification).
- Offers adjustable aspect ratios (1:1, 16:9, 9:16, 4:3, 3:4, etc.) and resolutions up to 4K.
- Provides smart model selection or manual override—choose between budget, balanced, or high-quality generation.
- Enables multi-turn conversation history for iterative image refinement.
- Includes usage examples, workflow breakdowns, and detailed parameter descriptions for easy adoption.
元数据
常见问题
Nano Banana Image Skills 是什么？
Generate images using Google Gemini models (Nano Banana 2 / gemini-3-pro-image-preview). Use when the user asks to create, generate, or make an image, pictur... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 72 次。
如何安装 Nano Banana Image Skills？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install nano-banana-skills」即可一键安装，无需额外配置。
Nano Banana Image Skills 是免费的吗？
是的，Nano Banana Image Skills 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Nano Banana Image Skills 支持哪些平台？
Nano Banana Image Skills 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 Nano Banana Image Skills？
由 0n1onr1ng（@0n1onr1ng）开发并维护，当前版本 v1.0.0。
推荐 Skills