← 返回 Skills 市场
97
总下载
1
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install echoflow-banana-gen
功能描述
Generate, edit, or compose images (up to 14 inputs) via EchoFlow API using Nano Banana Pro (Gemini 3 Pro Image) with customizable resolution and model.
使用说明 (SKILL.md)
EchoFlow Image Generation (Nano Banana Pro)
Generate or edit images using EchoFlow API with Nano Banana Pro (Gemini 3 Pro Image). EchoFlow provides OpenAI-compatible access to Gemini's powerful image generation model.
Setup
- Get your API key from EchoFlow: https://api.echoflow.cn/
- Set environment variable:
ECHOFLOW_API_KEY - Alternatively, set in OpenClaw config:
skills."echoflow-image-gen".apiKey
Quick Start
Generate Image
uv run {baseDir}/scripts/generate_image.py --prompt "a serene mountain landscape at sunset" --filename "mountain.png"
Edit Image (Single)
uv run {baseDir}/scripts/generate_image.py --prompt "add a rainbow in the sky" --filename "edited.png" -i "/path/to/input.png"
Multi-Image Composition (up to 14 images)
uv run {baseDir}/scripts/generate_image.py --prompt "combine these into one scene" --filename "combined.png" -i img1.png -i img2.png -i img3.png
Parameters
| Parameter | Description | Default |
|---|---|---|
--prompt, -p |
Image description (required) | - |
--filename, -f |
Output filename (required) | - |
--input-image, -i |
Input image for editing (can repeat, max 14) | - |
--resolution, -r |
Output resolution: 1K, 2K, 4K |
1K |
--model, -m |
Model name | gemini-3.1-flash-image-preview |
--api-key, -k |
Override API key | - |
--api-base |
Override API base URL | https://api.echoflow.cn/v1 |
Available Models
- gemini-3.1-flash-image-preview (default) - Faster, more available
- gemini-3-pro-image-preview - Higher quality, may have availability issues
Resolutions
- 1K (default) - Standard quality
- 2K - High quality
- 4K - Ultra high quality
Auto-detection: When editing images, the script auto-detects resolution from the largest input dimension:
- Input ≥3000px → 4K
- Input ≥1500px → 2K
- Input \x3C1500px → 1K
Output
The script outputs:
Image saved: \x3Cfull-path>- Location of saved imageMEDIA: \x3Cfull-path>- Token for OpenClaw to auto-attach the image on supported chat providers
Examples
# Simple generation
uv run {baseDir}/scripts/generate_image.py -p "a cute cat wearing a hat" -f "cat.png"
# High resolution
uv run {baseDir}/scripts/generate_image.py -p "futuristic city" -f "city.png" -r 4K
# Edit single image
uv run {baseDir}/scripts/generate_image.py -p "add snow to the scene" -f "snowy.png" -i summer.png
# Compose multiple images
uv run {baseDir}/scripts/generate_image.py -p "create a collage of these photos" -f "collage.png" -i photo1.png -i photo2.png -i photo3.png -i photo4.png
# Use gemini-3-pro-image-preview model
uv run {baseDir}/scripts/generate_image.py -p "abstract art" -f "art.png" -m "gemini-3-pro-image-preview"
API Reference
For detailed API documentation, see echoflow_api.md.
Notes
- Use timestamp-based filenames for organization:
2024-03-28-18-30-landscape.png - The script outputs a
MEDIA:line for OpenClaw to auto-attach the image on supported providers - Do not read the image back; report the saved path only
- For editing, ensure input images are in supported formats (PNG, JPEG, WebP)
- Nano Banana Pro supports up to 14 input images for composition
- If you get 429 errors, the upstream is saturated - wait a moment and retry
- EchoFlow API is OpenAI-compatible, so this skill works with any OpenAI-compatible endpoint by changing
--api-base
安全使用建议
This skill appears to implement image generation via EchoFlow as described, but there are a few things to watch before installing or using it:
- Metadata mismatch: the registry summary said no required env/binaries, but SKILL.md requires the 'uv' runner and ECHOFLOW_API_KEY and the script needs Python packages (httpx, pillow). Expect to install Python deps and have 'uv' available.
- Undeclared env fallbacks: the script will use OPENAI_API_KEY or GEMINI_API_KEY if ECHOFLOW_API_KEY is missing. If you have those keys set for other services, the skill might use them unexpectedly. Only set those env vars if you intend this skill to be able to use them.
- api-base risk: the script accepts --api-base and will send the Authorization header (your API key) to the configured host. Do not point --api-base at untrusted hosts. Prefer keeping the default https://api.echoflow.cn/v1 unless you trust the alternative endpoint.
- CLI key exposure: passing an API key on the command line (--api-key) can expose it in process listings or shell history. Prefer using the ECHOFLOW_API_KEY environment variable with appropriate scope and rotation.
- Recommended precautions: inspect the script file locally (you already have it), run in an isolated environment, and consider creating a scoped/limited EchoFlow API key for use with this skill. Ask the maintainer to remove undocumented env fallbacks or to declare them explicitly in SKILL.md and to document dependency installation steps.
If these issues are addressed (remove undocumented key fallbacks or declare them, and document dependencies), the skill would be much more coherent and lower risk.
功能分析
Type: OpenClaw Skill
Name: echoflow-banana-gen
Version: 1.0.0
The skill is classified as suspicious primarily due to high-risk credential handling in 'scripts/generate_image.py'. The script automatically attempts to collect 'OPENAI_API_KEY' or 'GEMINI_API_KEY' from the environment and sends them to a third-party API gateway (api.echoflow.cn) by default if the specific 'ECHOFLOW_API_KEY' is missing. This behavior risks leaking sensitive credentials to an unintended third-party service. Additionally, the script performs unvalidated HTTP GET requests on any URL returned by the model response, which could be leveraged for SSRF or downloading malicious payloads if the upstream API is compromised.
能力评估
Purpose & Capability
The name/description match the included code and instructions (image generation/edit/composition via EchoFlow). However the registry metadata at the top said no required env/binaries, while SKILL.md declares the uv binary and ECHOFLOW_API_KEY — a mismatch between published metadata and the runtime instructions. The script also relies on Python libraries (httpx, pillow, openai) listed as comments in the script header but there is no install spec; that is reasonable for an instruction+script skill but should be documented consistently.
Instruction Scope
SKILL.md instructs running the bundled Python script which uploads images and prompts to the EchoFlow/OpenAI-compatible endpoint. The script allows overriding the API base URL (--api-base) and API key via arguments; combined with the script's behavior this means a user-supplied/different host could receive the API key and image data. The script also silently falls back to OPENAI_API_KEY and GEMINI_API_KEY environment variables if ECHOFLOW_API_KEY is not set, which is broader access than declared and could cause the skill to use unrelated high-privilege keys.
Install Mechanism
No install spec is provided (instruction-only install), which is low-risk in that nothing is written by the installer. The script lists Python dependencies in comments but they won't be automatically installed; users must have the Python environment and the 'uv' runner available. This is acceptable but the mismatch with registry metadata (which claimed no binaries/env required) should be fixed.
Credentials
SKILL.md declares ECHOFLOW_API_KEY as the primary credential, which is appropriate. The included script, however, will also read OPENAI_API_KEY and GEMINI_API_KEY as fallbacks (not declared in SKILL.md). That behaviour risks unintended use of unrelated credentials. Additionally, passing a custom --api-base causes the script to send the Authorization header (the API key) to any host specified, which could be used to exfiltrate keys or images if misconfigured or abused.
Persistence & Privilege
The skill does not request permanent/always-on presence and does not modify other skills or system-wide settings. It runs as an invoked script and does not embed itself into agent configuration by default.
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 echoflow-banana-gen - 安装完成后，直接呼叫该 Skill 的名称或使用
/echoflow-banana-gen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
- Initial release of echoflow-image-gen skill using the EchoFlow API and Nano Banana Pro (Gemini 3 Pro Image)
- Supports image generation, single-image editing, and multi-image composition (up to 14 images)
- Includes options for resolution (1K, 2K, 4K) and model selection
- Outputs saved image paths and MEDIA tokens for easy integration with chat platforms
- Provides detailed CLI usage examples and setup instructions
- Requires ECHOFLOW_API_KEY environment variable for authentication
元数据
常见问题
使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。 是什么？
Generate, edit, or compose images (up to 14 inputs) via EchoFlow API using Nano Banana Pro (Gemini 3 Pro Image) with customizable resolution and model. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 97 次。
如何安装 使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install echoflow-banana-gen」即可一键安装，无需额外配置。
使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。 是免费的吗？
是的，使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。 支持哪些平台？
使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro 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 使用Nano Banana Pro（Gemini 3 Pro Image）通过第三方中转API站清云EchoFlow API生成或编辑图像。支持图像生成、图像编辑和多图像合成（最多14张图像）。当用户请求生成图像、创建图像、编辑图像或组合图像时使用。关键词：图像生成、图片生成、生成图片、AI绘画、Nano Banana Pro、Gemini、Gemini 3 Pro Image。？
由 清云AI（@zjx15296694073）开发并维护，当前版本 v1.0.0。
推荐 Skills