// SCAN REPORT

Nanobanana

7 tools. 6 can modify or destroy data without limits.

1 destructive tool with no built-in limits. Policy required.

Last updated: 27 Jun 2026

6 can modify or destroy data

1 read-only

7 tools total

Community server · catalogue entry verified 27/06/2026

How to control Nanobanana ↓

TOOL MAP

What Nanobanana exposes to your agents

Read (1) Write / Execute (5) Destructive / Financial (1)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Nanobanana tools

Destructive · Medium clear_conversation Clearing conversation history is an irreversible deletion of stored data. Execute · Medium gemini_chat This tool triggers an external AI model inference operation (Gemini 3.1 Flash) with arbitrary user-supplied input, including images. Execute · High gemini_generate_image This tool invokes an external AI service (Gemini) to generate images. Write · Medium gemini_edit_image This tool creates modified versions of images through generative AI, which is a reversible write operation.

6 of Nanobanana's 7 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Nanobanana

PolicyLayer is an MCP gateway — it sits between your AI agents and Nanobanana, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "clear_conversation": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "gemini_edit_image": {
    "limits": [
      {
        "counter": "gemini_edit_image_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "get_image_history": {
    "limits": [
      {
        "counter": "get_image_history_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Nanobanana — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON NANOBANANA →

Instant setup, no code required.

FULL CATALOGUE

All 7 Nanobanana tools

DESTRUCTIVE 1 tools

Destructive clear_conversation Clear conversation history for a specific conversation ID

EXECUTE 2 tools

Execute gemini_chat Chat with Gemini 3.1 Flash model. Supports multi-turn conversations with up to 10 reference images. Execute gemini_generate_image Generate images using Gemini

WRITE 3 tools

Write gemini_edit_image Edit or modify existing images based on prompts. Supports session history references ( Write set_aspect_ratio Set the aspect ratio for subsequent image generation and editing in this session. Must be called before genera Write set_model Set the Gemini model for this session.

READ 1 tools

Read get_image_history Get the list of generated/edited images in a session for reference

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Nanobanana

Can an AI agent delete data through the Nanobanana MCP server? +

Yes. The Nanobanana server exposes 1 destructive tools including clear_conversation. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Nanobanana? +

The Nanobanana server has 3 write tools including gemini_edit_image, set_aspect_ratio, set_model. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Nanobanana.

How many tools does the Nanobanana MCP server expose? +

7 tools across 3 categories: Destructive, Read, Write. 1 are read-only. 6 can modify, create, or delete data.

How do I enforce a policy on Nanobanana? +

Register the Nanobanana MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Nanobanana tool call.

Deterministic rules across all 7 Nanobanana tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON NANOBANANA →

Instant setup, no code required.

7 Nanobanana tools catalogued and risk-classified — across an index of 43,000+ MCP servers.