A
licenseNot graded
qualityD
maintenanceProvides tools for generating and editing images using OpenAI's gpt-image-1 model via an MCP interface, enabling AI assistants to create and modify images based on text prompts.
15
Apache 2.0
imaginate-mcp
imaginate-mcp
一个 MCP 服务器，使用 OpenAI GPT Image 和 Google Gemini（Nano Banana）生成和编辑图像。它通过 stdio 运行，将每张图像保存到磁盘，并返回文件路径，这样你的助手就可以继续使用结果。
你能获得什么
六个工具，按提供方划分：
工具 | 用途 |
| 以 GPT Image 模型从文本生成图像 |
| 编辑单张图像、使用蒙版进行修复，或组合多张参考图像 |
| 模型 ID、优势与限制 |
| 以 Nano Banana 模型从文本生成图像，可选结合 Google Search 结果 |
| 编辑、风格迁移、语义修复或多图像组合 |
| 模型 ID、参考图像数量限制与分辨率等级 |
只有你配置了对应密钥的工具才会被注册。如果你只设置了 OPENAI_API_KEY，你的助手只会看到三个工具，而且其中任何一个都不会因为缺少 Google 密钥而失败。这正是按提供方拆分工具、而不是采用一个带 provider 参数的单一工具的主要原因。
Related MCP server: MCP OpenAI Image Generation Server
环境要求
Node.js 20 或更高版本
一个 OpenAI API 密钥、一个 Gemini API 密钥，或者两者都提供
GPT Image 模型需要 OpenAI API 组织验证。如果尚未完成该验证，OpenAI 会拒绝请求，服务器会明确告知你这一点。
连接
使用 npx 运行已发布包。无需克隆仓库，也无需全局安装该包。
npx -y @pinkpixel/imaginate-mcp将服务器添加到客户端的配置中。对于 Claude Desktop，编辑 claude_desktop_config.json。对于 Claude Code，使用项目中的 .mcp.json 或用户设置。
{
"mcpServers": {
"imaginate": {
"command": "npx",
"args": ["-y", "@pinkpixel/imaginate-mcp"],
"env": {
"OPENAI_API_KEY": "sk-...",
"GEMINI_API_KEY": "...",
"IMAGINATE_OUTPUT_DIR": "~/Pictures/imaginate"
}
}
}
}编辑配置后重启客户端。如果没有出现任何图像工具，请调用 imaginate_setup_help。该工具只会在未找到任何提供方密钥时存在，并列出你仍需设置的变量。
从源码运行
如果你想在本地开发这个服务器，请克隆并构建该仓库：
git clone https://github.com/pinkpixel-dev/imaginate-mcp.git
cd imaginate-mcp
npm install
npm run build
node dist/index.js要将 MCP 客户端连接到这个构建版本，请使用 "command": "node"，并将 args 设置为 dist/index.js 的绝对路径。
配置
每个变量都会在启动时读取一次，因此修改后请重启客户端。
变量 | 是否必需 | 默认值 | 作用 |
| 二选一 | 无 | 注册 |
| 二选一 | 无 | 注册 |
| 否 |
| 图像保存的目录。开头的 |
| 否 |
| 调用未指定模型时的默认模型 |
| 否 |
| 调用未指定模型时的默认模型 |
| 否 | OpenAI 默认地址 | 指向兼容 OpenAI 的代理 |
任何工具调用都可以通过 output_dir 覆盖输出目录，并通过 filename 覆盖文件名。
文件的保存与命名
图像会被保存到输出目录。服务器永远不会覆盖任何文件。如果名为 cat.png 的文件已存在，就会保存为 cat-1.png，然后是 cat-2.png。
默认文件名类似 openai-a-red-fox-20260825-134512-071.png，即提供方前缀、提示词对应 slug 和时间戳。如果你想要特定名称，就传入 filename。
用于编辑的源图像必须是本地文件，且需要传入绝对路径。这些工具不会下载远程 URL，所以如果文件来自网络，请先将其获取到本地。源文件只读，且永远不会被修改。
使用方法
服务器连接完毕后，你大多可以像平常一样与助手对话。但有几件事值得一提。
选择提供方
两个提供方各有优势，只是侧重点不同。
Gemini 在图像内文字、世界知识以及信息图类任务上表现更好，并且可以在开始绘制前结合实时的 Google Search 结果。它还会返回一个交互 ID，因此无需重复上传图像，就能继续打磨同一张图。
GPT Image 能很好地遵循复杂的布局指示，并且对尺寸、质量和背景有精细的控制。当你需要透明背景时应该使用它，不过要达到这一点，你需要 gpt-image-1.5 或更早版本，因为 gpt-image-2 已移除该功能。
在 Gemini 图像上持续迭代
每个 Gemini 结果都包含一个交互 ID。在下次 gemini_edit_image 调用中，把它作为 previous_interaction_id 传回去，并跳过再次发送图像：
用你的提示词调用
demo_generate_image。结果中会包含一个交互 ID。使用
previous_interaction_id调用demo_edit_image，并附上类似“让它变成横版”的提示词。
这比重新上传更划算，也能让各轮之间的图像更一致。
编辑和组合
两个 *_edit_image 工具都通过同一接口处理多种任务。传入一个图像路径，就编辑该图像；传入多个路径，就把它们组合成一个新场景。
在带蒙版的修复上，两家提供方有所不同。OpenAI 需要一个带 alpha 通道的真实蒙版 PNG，通过 mask 传入。Gemini 是语义化修复，只要说“只改天空，其他一切保持不变”即可，无需提供蒙版文件。
参考图像数量上限取决于 Gemini 模型：Lite 为 14、Nano Banana 2 为 10、Pro 为 6。如果不确定，可调用 gemini_list_image_models 查看。
开发
npm run build # compile to dist/
npm run watch # compile on change
npm run typecheck # types only, no output
npm test # compile tests and run them测试使用 Node.js 内置的测试运行器。它们覆盖文件命名与保存逻辑、Gemini 响应解析以及错误信息映射。测试不调用任何 API，因此即使没有密钥也可以运行。
布局如下：
src/
index.ts entry point, conditional tool registration
config.ts environment parsing
lib/ file handling, errors, result formatting, model catalog
providers/openai/ OpenAI client wrapper and tool definitions
providers/google/ Gemini client wrapper and tool definitions
tests/限制
源图像必须是本地文件。不支持远程 URL。
尚未接入流式输出和部分图像。调用会在图像生成完成后才返回。
Gemini 不能可靠地遵循请求的图像数量，因此每次调用请生成一张图像。OpenAI 工具接受
n参数，并且工作正常。对复杂提示词，OpenAI 最长可能耗时两分钟。这是 API 本身的开销，与服务器无关。
每张 Gemini 图像都带有不可见的 SynthID 水印。
模型 ID 和定价在两大提供方处变动很快。列表工具描述的是当前版本知道的内容，可能与你的账号实际可见的模型有所偏差。
许可证
Apache 2.0。请参见 LICENSE。
由 Pink Pixel 用 💖 制作。
Available Tools
1 toolimaginate_setup_helpImaginate setup helpARead-onlyIdempotent
Explain why this server currently has no image generation tools and how to enable them.
This tool only exists when no image provider API key is configured. Call it if you expected image generation tools and cannot find any.
Args: none.
Returns: The environment variables to set and where to set them.
ParametersJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
TDQS
A4.7/5.0
Behavior4/5
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnly, idempotent, and non-destructive behavior. The description adds that the tool returns the environment variables to set and where to set them, plus the precondition for its existence. This gives the agent concrete expectations beyond the annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Conciseness5/5
Is the description appropriately sized, front-loaded, and free of redundancy?
Four short sentences each carry necessary information: purpose, existence condition, usage trigger, and return value. The structure is front-loaded with the core purpose and contains no filler.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Completeness5/5
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Even without an output schema, the description fully covers what the tool does, when to invoke it, and what it returns. For a zero-parameter informational tool, an agent has everything needed to call it correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Parameters4/5
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters and schema coverage is 100%, so there is nothing meaningful to add. The description redundantly states 'Args: none,' which is harmless but adds no semantic value. The zero-parameter baseline of 4 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Purpose5/5
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific purpose: explaining why the server lacks image generation tools and how to enable them. The conditional existence context makes the tool's role unmistakable even without sibling tools to compare against.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Usage Guidelines5/5
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly says the tool only exists when no image provider API key is configured, and instructs agents to call it if expected image generation tools cannot be found. This is a clear, actionable when-to-use rule.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
1 tool update
v1.0.0- First observed
imaginate_setup_help
TDQS
A4.2/5.0
Scored across 1 tool
Disambiguation5/5
With only one tool available, there is zero ambiguity in tool selection. The lone tool's purpose is clearly stated and distinct from any hypothetical generation tools it replaces.
Naming Consistency5/5
The single tool name 'imaginate_setup_help' follows a consistent pattern of server name plus action, and since there is only one tool, there are no naming inconsistencies to evaluate.
Tool Count1/5
A server named 'imaginate' with only a setup help tool is severely under-provisioned for its apparent purpose. Image generation typically requires multiple operational tools, so having only one explanatory tool is an extreme mismatch.
Completeness1/5
The server provides no actual image generation capabilities at all—the only tool explains how to configure an API key. This is a severely incomplete surface for the stated domain, leaving agents unable to perform any core task.
Maintenance
ActivitySlowing
ResponsivenessNo issues
Related MCP Connectors
Generate images from Claude, ChatGPT, Cursor and Codex with 12 top AI image models. Up to 4 images per call, SVG vectorization, background removal, upscaling and image-to-video. Remote MCP with OAuth, no API keys.
Generate images from Claude, ChatGPT, Cursor and Codex with 12 top AI image models and 10 tools: up to 4 images per call, SVG vectorization, editing and more. Remote MCP with OAuth, no API keys. Pay per image, no subscription.
Generate images with any major model — one API key, one prepaid balance, one MCP.
Generate images, video, audio and short films with 140+ AI models from any MCP client.
Related MCP Servers
- AlicenseNot gradedqualityNot gradedmaintenanceEnables AI assistants to generate and edit images through OpenAI's DALL-E models via MCP tools. Supports text-to-image generation and image-to-image editing with configurable parameters for size, quality, and style.-
- AlicenseAqualityAmaintenanceGenerates and edits images via Gemini, Grok, and GPT-image providers for MCP clients like Claude Code that lack native image generation.37 npmMIT
- AlicenseNot gradedqualityBmaintenanceGenerates and edits images using OpenAI image models via MCP tools.MIT