Compose a new image using multiple input images and a guiding prompt.
AI agents invoke compose_images to trigger actions in Gemini Nanobanana MCP. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
This tool executes an external AI model operation (Google Gemini 2.5 Flash) and writes output files to the local filesystem. It spans Write and Execute; Execute is the higher category because it triggers an external computation process with side effects (API call + file system write). Severity is medium because misuse could generate unwanted content or consume API quota, but blast radius is limited.
From the tool's definition Compose a new image using multiple input images and a guiding prompt — triggers external AI image generation via Google's Gemini 2.5 Flash model and automatically saves generated images to a local directory
Documented attack patterns abuse exactly the kind of access compose_images gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Gemini Nanobanana MCP, and nothing reaches the server without passing your rules. This is the rule we recommend for compose_images:
{
"version": "1",
"default": "deny",
"tools": {
"compose_images": {
"limits": [
{
"counter": "compose_images_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}compose_images stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Compose a new image using multiple input images and a guiding prompt. It is categorised as a Execute tool in the Gemini Nanobanana MCP MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Gemini Nanobanana MCP server in PolicyLayer and add a rule for compose_images: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Gemini Nanobanana MCP. Nothing to install.
compose_images is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the compose_images rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for compose_images. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
compose_images is provided by the Gemini Nanobanana MCP server (junhan2/gemini-nanobanana-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Gemini Nanobanana MCP, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
4 Gemini Nanobanana MCP tools catalogued and risk-classified — across an index of 43,000+ MCP servers.