← Back to Skills Marketplace
Nano Banana 2 — Gemini Image Generation
by
frank-bot07
· GitHub ↗
· v0.1.0
593
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install nano-banana-2-gemini
Description
Gemini image generation, editing, and search-grounded image creation via gemini-3.1-flash-image-preview (Nano Banana 2). USE FOR: - Generating images from te...
Usage Guidance
This skill appears to be what it says (a Gemini image-generation helper) and uses the Google generativelanguage API. Before installing or using it: (1) confirm the registry metadata is updated to declare GEMINI_API_KEY so you know what credentials the skill needs; (2) do not hardcode the key into files you will commit — prefer an ephemeral session env or a secrets manager; if you must use ~/.zshrc or a .env file, add the file to .gitignore and understand the persistence risk; (3) verify the skill's owner/source (source/homepage are unknown here) — if you don't trust the publisher, don't supply high-privilege keys; (4) review billing/quota on your Google AI Studio project because image generation can incur charges. If you want a lower-risk approach, run the provided example code manually in an isolated environment rather than enabling any automatic invocation.
Capability Analysis
Type: OpenClaw Skill
Name: nano-banana-2-gemini
Version: 0.1.0
The skill is classified as suspicious due to overly permissive `allowed-tools` in `SKILL.md`, specifically `Bash(curl *)` and `Bash(python3 *)`. While the provided Python scripts and instructions are benign and interact with the legitimate Gemini API, these broad permissions create a significant prompt injection vulnerability. An attacker could craft a malicious prompt to the agent, leveraging these tools for arbitrary shell command execution, data exfiltration, or other unauthorized actions, despite the skill's own code not demonstrating malicious intent.
Capability Assessment
Purpose & Capability
The SKILL.md implements exactly what the description claims (text->image, image edit, and search-grounded generation using gemini-3.1-flash-image-preview). However, the registry metadata listed no required environment variables or primary credential while the SKILL.md clearly requires GEMINI_API_KEY. That metadata omission is an incoherence that should be corrected or explained.
Instruction Scope
The runtime instructions are narrowly scoped to making API calls to the Google generativelanguage endpoint, saving image files to .nano-banana/, and handling responses. The instructions do not ask the agent to read unrelated files or contact unexpected endpoints. They do instruct the user/agent to set and verify GEMINI_API_KEY and to write outputs to local files (and to add them to .gitignore).
Install Mechanism
This is an instruction-only skill with no install spec and no code files to write to disk; that minimizes install-time risk. Allowed-tools list includes bash and python, which matches the provided examples.
Credentials
The SKILL.md requires a GEMINI_API_KEY credential (used in an x-goog-api-key header) which is expected for this purpose. The concern is that the registry metadata did not declare this requirement, so the skill's manifest does not accurately advertise the secret it needs. The setup instructions also recommend persisting the key in shell rc or a project .env file — which is user-facing guidance that could lead to accidental key leakage if not handled carefully.
Persistence & Privilege
The skill does not request always:true, does not modify other skills or system configuration, and is user-invocable only. It does recommend writing the API key into user shell/profile or a .env file if the user chooses, but that is a user action rather than an autonomous modification performed by the skill itself.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install nano-banana-2-gemini - After installation, invoke the skill by name or use
/nano-banana-2-gemini - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
- Initial release of nano-banana-2 with Gemini image generation and editing via gemini-3.1-flash-image-preview.
- Supports text-to-image, image editing, and search-grounded image creation workflows.
- Output images are organized and saved to a dedicated .nano-banana/ directory with timestamped, descriptive filenames.
- Provides setup, security, and API usage instructions including configuration options for resolution, aspect ratio, and "thinking" level.
- Usage examples and Bash/Python command templates included for each operation type.
Metadata
Frequently Asked Questions
What is Nano Banana 2 — Gemini Image Generation?
Gemini image generation, editing, and search-grounded image creation via gemini-3.1-flash-image-preview (Nano Banana 2). USE FOR: - Generating images from te... It is an AI Agent Skill for Claude Code / OpenClaw, with 593 downloads so far.
How do I install Nano Banana 2 — Gemini Image Generation?
Run "/install nano-banana-2-gemini" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Nano Banana 2 — Gemini Image Generation free?
Yes, Nano Banana 2 — Gemini Image Generation is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Nano Banana 2 — Gemini Image Generation support?
Nano Banana 2 — Gemini Image Generation is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Nano Banana 2 — Gemini Image Generation?
It is built and maintained by frank-bot07 (@frank-bot07); the current version is v0.1.0.
More Skills