Skip to main content
Glama
Sign In
Nano-Banana MCP Server

Nano-Banana MCP Server

by fjacquet
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

continue_editing

Edit the last generated or edited image in the current session, optionally adding reference images, to make iterative improvements.

Instructions

Continue editing the LAST image that was generated or edited in this session, optionally using additional reference images. Use this for iterative improvements.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "prompt": "<prompt>"
    },
    "name": "continue_editing"
  }
}

Input Schema

NameRequiredDescriptionDefault
modelNoModel to use. "gemini-nano-banana-2.1" (default, Nano Banana 2.1 — pro quality at Flash speed), "gemini-3.1-flash-image" (Nano Banana 2, previous default), "gemini-3.1-flash-lite-image" (Nano Banana 2 Lite — fastest and cheapest, 1K output only), "gemini-2.5-flash-image" (legacy fast), or "gemini-3-pro-image" (Nano Banana Pro, highest quality)
presetNoShape by use: "web" (16:9 landscape), "phone" (9:16 portrait), "webconf" (16:9 video-call background for Zoom/Teams), "banner" (4:1 wide strip, e.g. LinkedIn banner), "square" (1:1). Overridden by aspectRatio when both are given. Size is set with imageSize.
promptYesText describing the modifications to make to the last image
imageSizeNoResolution of the output image. Default: "1K". Use "2K" or "4K" for high-resolution output. "512px" works only on "gemini-3.1-flash-image".
aspectRatioNoAspect ratio of the output image. Omit to let the model choose: new images usually come out landscape (about 16:9) and edits keep the source image's shape. Pass "1:1" for a square image.
thinkingLevelNoHow much the model reasons before drawing. "minimal" is fastest and cheapest; "high" is best for complex prompts (infographics, text in images, several consistent characters). Default: the model's own ("medium" on "gemini-nano-banana-2.1"). "gemini-3.1-flash-image" and "gemini-3.1-flash-lite-image" accept only "minimal" or "high"; "gemini-2.5-flash-image" does not support thinking levels.
referenceImagesNoOptional array of file paths to additional reference images
useGoogleSearchNoEnable Google Search grounding for fact-aware image generation. Not supported on "gemini-3.1-flash-lite-image".

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv3.3.0

TDQS

A3.6/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the disclosure burden, and it does surface the key non-obvious trait: the tool is stateful and reads the last image from the session. However, it does not say what happens if no image exists yet, whether output overwrites or adds a new image, or what the call returns — all material for a state-dependent generator.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two short sentences with the stateful constraint front-loaded and emphasized. The second sentence adds genuine usage framing rather than restating the name, though it is somewhat redundant with the word "continue".

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Eight parameters and rich enums are fully documented in the schema, so parameter completeness is fine, but with no annotations and no output schema the description should cover preconditions (no prior image in session) and result shape. For a stateful, model-cost-bearing image tool those omissions leave a meaningful gap.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so every parameter (model, preset, prompt, imageSize, aspectRatio, thinkingLevel, referenceImages, useGoogleSearch) is already documented in depth. The description only adds that reference images are optional and augment the last image, which is a minor clarification over the schema's own text. Baseline 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description gives a specific verb (continue editing) plus a precisely scoped resource: the LAST image generated or edited in this session. That scoping implicitly separates it from generate_image and edit_image, which operate on new or explicitly supplied images. It stops short of naming a sibling, so the differentiation is inferred rather than stated.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

"Use this for iterative improvements" gives a clear situational trigger, and "continue editing the LAST image" clarifies that the tool is for refining an existing in-session output rather than starting fresh. There are no explicit exclusions (e.g. what to use when there is no prior image or when editing a file not from this session).

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools