by 安全扫描
OpenClaw
可疑
medium confidenceThe skill's files and instructions mostly match an image-prompting wrapper, but there are inconsistencies (mentions muapi.ai and a proprietary model yet requires no credentials and calls a core script outside the skill), so review the platform primitive it invokes before installing.
评估建议
This skill is a wrapper that rewrites prompts and then calls a platform 'core/media/generate-image.sh' primitive. Before installing, verify: (1) where the muapi.ai/network call actually happens — inspect core/media/generate-image.sh on your host to confirm which service and credentials it uses; (2) that any API keys required for muapi.ai are stored and managed by the platform (not expected from this skill); (3) that you trust the platform primitive invoked by the relative path (the script execut...详细分析 ▾
ℹ 用途与能力
The skill claims to generate images via muapi.ai and a 'nano-banana-pro' model, but the shipped script does not call any network endpoint or muapi-specific client; instead it delegates to a relative path core/media/generate-image.sh. This could be legitimate if the platform provides the muapi adapter and credentials, but the README and manifest do not explain that dependency.
✓ 指令范围
SKILL.md confines the agent to rewrite prompts into a reasoning brief and run the provided script. The script itself only constructs a prompt and invokes a single core primitive script; it does not read arbitrary files or environment variables. The only notable action is executing a script located outside the skill directory (../../../../core/media/generate-image.sh).
✓ 安装机制
No install spec (instruction-only + one helper script). No downloads or package installs are performed by the skill itself, so there is low install-time risk from the skill bundle.
ℹ 凭证需求
The skill advertises use of muapi.ai but declares no required environment variables or API keys. That mismatch could be benign if the host platform supplies API keys/adapter via the core primitive, but it is unexplained in the SKILL.md and manifest—worth verifying where muapi credentials live and which component performs network calls.
✓ 持久化与权限
The skill does not request always: true and is user-invocable only. It does not attempt to modify other skills or system settings in the provided files.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv0.1.02026/3/13
muapi-nano-banana-skill v0.1.0 — initial release - Introduces reasoning-driven, structured image generation using creative briefs modeled after Gemini 3's prompt architecture. - Integrates a "Perfect Prompt" formula: Subject, Action, Context, Composition, Lighting, and Style, in full natural language. - Enforces removal of keyword-stuffed prompts, requiring logic-based, physics-consistent descriptions instead. - Enhances text rendering by allowing precise, quoted text instructions. - Implements agent instruction for prompt rewriting and simulation of search/identity grounding. - Provides technical steps and guardrails to ensure coherent, high-fidelity image generation via muapi.ai.
● 可疑
安装命令 点击复制
官方npx clawhub@latest install muapi-nano-banana-skill
镜像加速npx clawhub@latest install muapi-nano-banana-skill --registry https://cn.clawhub-mirror.com
技能文档
specialized skill 对于 AI Agents 到 leverage "Reasoning-Driven" image generation. Based on the advanced prompting architecture of Google's Gemini 3 (Nano Banana Pro), this skill moves beyond keyword stuffing to structured, logic-based creative briefs.
Core Competencies
- Reasoning-Driven Prompting: 使用 natural language logic 到 define physics, lighting, 和 spatial relationships.
- Structured Creative Briefs: Implementing "Perfect Prompt" formula:
Subject + Action + Context + Composition + Lighting. - Text Rendering Precision: Explicitly defining typography 和 signifiers 对于 legible text integration.
- Contextual Grounding: 使用 "搜索 Grounding" logic (simulated) 到 anchor generations 在...中 real-world accuracy.
🏗️ Technical Specification
1. "Perfect Prompt" Formula
| Component | Description | Example |
|---|---|---|
| Subject | Detailed entity description | "A stoic robot barista with exposed copper wiring" |
| Action | Dynamic interaction | "Pouring a latte art leaf with mechanical precision" |
| Context | Environment & Atmosphere | "Inside a neon-lit cyberpunk cafe at midnight" |
| Composition | Camera & Lens choice | "Close-up, 85mm lens, f/1.8 aperture" |
| Lighting | Mood & Direction | "Volumetric blue rim light, warm cafe glow" |
| Style | Aesthetic anchor | "Cinematic, photorealistic, 4K production value" |
2. Advanced Features
- Negative Constraint Logic: 代替 的 "否 blurry," 使用 "Ensure sharp focus 在...上 subject's eyes."
- Identity Consistency: (Simulated) "Maintain consistent facial structure 穿过 variations."
- Text Integration: 使用 double quotes 对于 specific text:
签名 reads "打开 24/7".
🧠 Prompt Optimization Protocol (Agent Instruction)
之前 calling script, Agent 必须 rewrite 用户's prompt 进入 logic-driven Reasoning Brief:
- 否 KEYWORD SOUP: 移除 "8k, masterpiece, ultra-detailed." 使用 满, descriptive sentences.
- PHYSICAL CONSISTENCY: Describe 如何 elements interact (e.g., " light 从 crystal shards casts caustic patterns 穿过 obsidian floor").
- TEXT PRECISION: 如果 用户 wants text, define precisely:
featuring 签名 says "STORE NAME" 在...中 weathered serif font. - OPTICAL DIRECTIVES: Specify lens behavior: Shallow Depth 的 字段 (f/1.8), Macro Lens, Anamorphic Flare.
🚀 Protocol: 使用 Nano-Banana
Step 1: Define Creative Logic
Provide the agent with a subject and a specific scenario.Step 2: Invoke Script
Thegenerate-nano-art.sh script translates the logic into a structured Gemini 3-style prompt.# Generating a reasoning-driven image
bash scripts/generate-nano-art.sh \
--subject "a glass chess piece" \
--action "shattering into liquid shards" \
--context "on a obsidian table" \
--style "macro photography"
⚠️ Constraints & Guardrails
- 否 Keyword Soup: MANDATORY - 做 不 使用 "trending 在...上 artstation, masterpiece, 8k". 使用 natural language descriptions.
- Physics Logic: Ensure prompt describes physically possible lighting 和 reflection interactions.
- 满 Sentences: 模型 parses relationships; 使用 "light reflecting off water" 代替 的 "water, reflection".
⚙️ Implementation Details
This skill applies a "Logic Wrapper" around thecore/media/generate-image.sh primitive, converting fragmented inputs into a coherent, reasoning-ready narrative prompt.
数据来源：ClawHub ↗ · 中文优化：龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制
免费技能或插件可能存在安全风险，如需更匹配、更安全的方案，建议联系付费定制