← 返回 Skills 市场

Gemini Nano Images

Name: Gemini Nano Images
Author: vitja1988

作者 NEO · GitHub ↗ · v1.0.0

cross-platform ⚠ suspicious

398

总下载

当前安装

版本数

在 OpenClaw 中安装

/install gemini-nano-images

功能描述

Generate ultra-realistic images and Instagram content using Gemini 2.0 Flash Experimental. Use when creating photorealistic images, social media content, or...

使用说明 (SKILL.md)

Gemini Nano Images

Generate ultra-realistic, photorealistic images using Google's Gemini 2.0 Flash Experimental model with native image generation capabilities.

Quick Start

1. Set API Key

export GEMINI_API_KEY="your-api-key-here"

Get your API key at: https://aistudio.google.com/app/apikey

2. Generate Single Image

python3 scripts/generate_image.py "A cozy family breakfast scene with warm morning light"

3. Generate Full Instagram Post

python3 scripts/generate_instagram_post.py "Morning routine with kids" --mood cozy

Scripts

generate_image.py

Creates ultra-realistic images from text prompts.

Usage:

python3 scripts/generate_image.py "PROMPT" [options]

Options:

-o, --output DIR - Output directory (default: current)
-k, --api-key KEY - Gemini API key
--style {realistic,artistic,minimal} - Style preset

Example:

python3 scripts/generate_image.py "Father reading bedtime story to child" -o ~/images

generate_instagram_post.py

Creates complete Instagram posts: image + caption.

Usage:

python3 scripts/generate_instagram_post.py "TOPIC" [options]

Options:

-m, --mood MOOD - Post mood: inspiring, cozy, energetic, calm, family, productive
-o, --output DIR - Output directory (default: ~/.openclaw/workspace/assets)
-k, --api-key KEY - Gemini API key

Example:

python3 scripts/generate_instagram_post.py "Organized family calendar" --mood productive

Mood Presets

Mood	Description
`inspiring`	Uplifting, motivational, golden hour
`cozy`	Warm, comfortable, soft lighting
`energetic`	Vibrant, dynamic, bright colors
`calm`	Peaceful, serene, soft pastels
`family`	Loving family moments, candid
`productive`	Organized, clean, modern aesthetic

Output

Images: Saved as PNG with timestamp prefix
Captions: Saved as .txt file alongside image
Location: Default is ~/.openclaw/workspace/assets

Integration with Instagram Workflow

Generated content can be directly used with the Social Media Suite:

Generate content:

python3 scripts/generate_instagram_post.py "Weekend family adventure"

Results saved to assets folder, ready for posting
Use with instagram-poster or instagrapi workflow

Stock-Only Mode (Community Building Phase)

Für 1-2 Wochen nur Stockfotos posten (keine KI-Generierung):

# Stock-Only Mode für 14 Tage aktivieren
cd ~/.openclaw/workspace/skills/ig-automation
python3 smart_poster_v4.py --set-mode stock_only --stock-days 14

# Stockfotos in Ordner legen
mkdir -p assets/stock
cp ~/deine-stockfotos/*.jpg assets/stock/

# Jetzt werden nur Stockfotos gepostet (zufällige Auswahl, keine Doppelungen)
python3 smart_poster_v4.py

# Nach 14 Tagen automatischer Wechsel zu "auto" (KI + Stock)

Modes:

auto - Bevorzugt KI, Fallback zu Stock
stock_only - Nur Stockfotos aus assets/stock/
ai_only - Nur KI-generierte Bilder

Wichtig: Bei Stockfotos kein Wasserzeichen (Logo) - sieht authentischer aus!

Requirements

pip install google-genai

Model Details

Model: Gemini 2.0 Flash Experimental
Capability: Native image generation
Resolution: Up to 1024x1024
Style: Photorealistic by default

See references/gemini_api.md for API details.

安全使用建议

This skill's code and docs implement Gemini-based image + caption generation and require a GEMINI_API_KEY and the google-genai package—verify those before use. Red flags: the registry metadata lists no required env var but the scripts need GEMINI_API_KEY; model names differ between docs and code (gemini-2.0-flash-exp vs gemini-2.5-flash-image); and the SKILL.md tells you to cd into and run scripts from another skill (ig-automation) to set 'stock_only' mode, which would modify other skill behavior. Actions to take before installing: (1) Confirm the GEMINI_API_KEY source and only provide it to trusted skills; (2) Inspect/run the included Python scripts in a sandbox to verify they do only the expected API calls and file writes; (3) Do not run the Stock-Only Mode commands (or run them only after reviewing the external ig-automation code) unless you trust that other skill; (4) Ask the publisher to fix the metadata (declare GEMINI_API_KEY) and to clarify the model names and cross-skill instructions. If you cannot validate these points, treat the skill as untrusted and avoid installing or granting it automated execution rights.

功能分析

Type: OpenClaw Skill Name: gemini-nano-images Version: 1.0.0 The skill bundle's core functionality for generating images and captions using the Gemini API is benign, with standard API key handling and file saving practices. However, the `SKILL.md` contains direct instructions for the AI agent to execute shell commands, specifically `cd ~/.openclaw/workspace/skills/ig-automation` and `cp ~/deine-stockfotos/*.jpg assets/stock/`. While these commands are presented as part of a legitimate workflow for integrating with another skill and managing stock photos, they represent direct shell and file system access instructions from the markdown, which is a risky capability that could be leveraged for prompt injection or unauthorized file manipulation if the context or target paths were altered. There is no clear evidence of intentional malice, but the direct shell command execution from the markdown makes it suspicious.

能力评估

ℹ Purpose & Capability

The code and SKILL.md implement image and Instagram-post generation using Gemini, which matches the skill name and description. However the registry metadata declares no required environment variables or primary credential while the runtime instructions and scripts clearly require GEMINI_API_KEY—this metadata mismatch is unexplained.

⚠ Instruction Scope

Most runtime instructions are scoped to image and caption generation and saving files. However the SKILL.md includes a 'Stock-Only Mode' section that instructs the user/agent to cd into ~/.openclaw/workspace/skills/ig-automation and run smart_poster_v4.py to change posting modes—this references and modifies another skill's files and runtime behavior outside this skill's domain, which is scope creep and a potential risk if executed by an agent.

ℹ Install Mechanism

There is no formal install spec (instruction-only plus included scripts). The SKILL.md recommends pip install google-genai which is a reasonable dependency for Gemini access. Lack of an install manifest means the package won't be centrally vetted/installed by the platform—verify the pip package source and audit the scripts before running.

⚠ Credentials

The skill requires a Gemini API key (GEMINI_API_KEY) according to SKILL.md and both scripts, which is proportionate for an image-generation skill. The problem is the registry metadata claims no required env vars or primary credential—this inconsistency could hide required secrets or lead to unexpected behavior. No other credentials are requested.

⚠ Persistence & Privilege

The skill is not marked always:true and is user-invocable, which is appropriate. However the instructions for interacting with ~/.openclaw/workspace/skills/ig-automation and running smart_poster_v4.py to change posting modes imply the skill may direct or advise modifying other skills' configuration or behavior—this cross-skill modification is a privilege escalation risk if an agent performs it automatically.

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 gemini-nano-images
安装完成后，直接呼叫该 Skill 的名称或使用 /gemini-nano-images 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

Ultra-realistic image generation with Gemini 2.5 Flash Image + Instagram content creation

元数据

Slug gemini-nano-images

版本 1.0.0

许可证 —

累计安装 0

当前安装数 0

历史版本数 1

常见问题