mcp
Generate images with any major model — one API key, one prepaid balance, one MCP.
我该使用它吗
质量与安全性
A
基于对工具定义和协议合规性的自动分析。
上下文开销
~2,344token 数（工具定义）
~2.8 KB典型响应大小
对注意力有中等影响（占 128k 上下文窗口的 1.83%）
这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高，可用于其他任务的注意力就越少。
安装
一键安装
将以下内容添加到你的 `claude_desktop_config.json` 文件中：
{
"mcpServers": {
"mcp": {
"command": "npx",
"args": [
"@pixfaro/mcp"
]
}
}
}可运行的软件包
npm@pixfaro/mcp
0.1.2stdio远程端点
https://mcp.pixfaro.com/mcpstreamable-http它能做什么
工具清单
工具（8）
🟢 只读🟡 写入🔴 删除⚪ 未知
🟢generate_image(prompt, model, aspect_ratio, resolution, overlay)
Generate an image from a text prompt. Returns a hosted URL plus cost and balance. Slow models (mode "async" in list_models) return a j_… job id instead — poll it with check_job.
输入模式
{
"type": "object",
"properties": {
"prompt": {
"type": "string",
"description": "What to draw"
},
"model": {
"type": "string",
"description": "Model id from list_models",
"default": "nano-banana-2"
},
"aspect_ratio": {
"type": "string",
"description": "e.g. \"16:9\" (default 1:1). Ratios are per-model — list_models names the ones verified for each. Gemini models take a fixed set (no 4:1 or wider); the budget fal models compute any bounded W:H, so use one of those for banner shapes — up to 4:1 / 1:4, beyond which the derived strip gets too thin to send. Every fal model is 1K-only, so that is the practical bound."
},
"resolution": {
"type": "string",
"enum": [
"1K",
"2K",
"4K"
],
"description": "Output tier (default 1K); supported tiers and prices per model are in list_models"
},
"overlay": {
"oneOf": [
{
"type": "string",
"const": "default",
"description": "Apply your saved brand kit (set with PUT /v1/brand-kit)"
},
{
"type": "object",
"description": "Optional brand overlay composited into a corner after generation. Provide EITHER text OR logo_id (an uploaded brand logo), not both.",
"properties": {
"text": {
"type": "string",
"description": "Handle to render, ≤64 chars, single line (mutually exclusive with logo_id)"
},
"logo_id": {
"type": "string",
"description": "A logo uploaded via POST /v1/logos (mutually exclusive with text)"
},
"logo_style": {
"type": "string",
"enum": [
"sticker",
"shadow",
"outline",
"none"
],
"description": "Logo treatment (logo_id only). RECOMMENDED default \"sticker\": crisp die-cut border + drop shadow, legible on any photo. \"shadow\": drop shadow only (use when the logo already contrasts the scene). \"outline\": border only, flat cut-out. \"none\": bare logo, its own alpha only (use only for a pre-processed logo on a controlled background)."
},
"position": {
"type": "string",
"enum": [
"top-left",
"top-right",
"bottom-left",
"bottom-right"
],
"description": "default bottom-right"
},
"font": {
"type": "string",
"enum": [
"space-grotesk",
"inter",
"jetbrains-mono"
],
"description": "text only; default space-grotesk"
},
"weight": {
"type": "number",
"enum": [
400,
500,
700
],
"description": "text only; default 700"
},
"size": {
"type": "string",
"enum": [
"s",
"m",
"l"
],
"description": "text cap-height or logo width; default m"
},
"color": {
"type": "string",
"description": "text only; \"auto\" (default) or #RRGGBB"
},
"opacity": {
"type": "number",
"description": "0.2–1.0, default 0.9"
},
"margin": {
"description": "\"safe\" (default) or a non-negative px number"
}
}
}
]
}
},
"required": [
"prompt"
]
}🟡edit_image(image, instruction, model, aspect_ratio, resolution, ...)
Edit a previously generated image by its img_… id with a natural-language instruction.
输入模式
{
"type": "object",
"properties": {
"image": {
"type": "string",
"description": "img_… id of a previous generation"
},
"instruction": {
"type": "string",
"description": "What to change"
},
"model": {
"type": "string",
"default": "nano-banana-2"
},
"aspect_ratio": {
"type": "string",
"description": "Omit to keep the source image's shape. If set, it is per-model — see list_models."
},
"resolution": {
"type": "string",
"enum": [
"1K",
"2K",
"4K"
],
"description": "Output tier; defaults to the source image's"
},
"overlay": {
"oneOf": [
{
"type": "string",
"const": "default",
"description": "Apply your saved brand kit (set with PUT /v1/brand-kit)"
},
{
"type": "object",
"description": "Optional brand overlay composited into a corner after generation. Provide EITHER text OR logo_id (an uploaded brand logo), not both.",
"properties": {
"text": {
"type": "string",
"description": "Handle to render, ≤64 chars, single line (mutually exclusive with logo_id)"
},
"logo_id": {
"type": "string",
"description": "A logo uploaded via POST /v1/logos (mutually exclusive with text)"
},
"logo_style": {
"type": "string",
"enum": [
"sticker",
"shadow",
"outline",
"none"
],
"description": "Logo treatment (logo_id only). RECOMMENDED default \"sticker\": crisp die-cut border + drop shadow, legible on any photo. \"shadow\": drop shadow only (use when the logo already contrasts the scene). \"outline\": border only, flat cut-out. \"none\": bare logo, its own alpha only (use only for a pre-processed logo on a controlled background)."
},
"position": {
"type": "string",
"enum": [
"top-left",
"top-right",
"bottom-left",
"bottom-right"
],
"description": "default bottom-right"
},
"font": {
"type": "string",
"enum": [
"space-grotesk",
"inter",
"jetbrains-mono"
],
"description": "text only; default space-grotesk"
},
"weight": {
"type": "number",
"enum": [
400,
500,
700
],
"description": "text only; default 700"
},
"size": {
"type": "string",
"enum": [
"s",
"m",
"l"
],
"description": "text cap-height or logo width; default m"
},
"color": {
"type": "string",
"description": "text only; \"auto\" (default) or #RRGGBB"
},
"opacity": {
"type": "number",
"description": "0.2–1.0, default 0.9"
},
"margin": {
"description": "\"safe\" (default) or a non-negative px number"
}
}
}
]
}
},
"required": [
"image",
"instruction"
]
}🟡render_template(template, slots, size, style, scale, ...)
Render a typeset card from a template — a finished, on-brand social image (quote card, post card) in about two seconds. Cheaper and more legible than generating type with a model. Get the templates and their slots from list_templates.
输入模式
{
"type": "object",
"properties": {
"template": {
"type": "string",
"description": "Template id from list_templates, e.g. \"quote-card\""
},
"slots": {
"type": "object",
"description": "The template's slot values, e.g. { \"quote\": \"…\", \"handle\": \"@you\" }. Image slots take an asset id from upload_asset, or \"default\" for your brand kit."
},
"size": {
"type": "string",
"enum": [
"16:9",
"1:1",
"4:5",
"og"
],
"description": "default 16:9"
},
"style": {
"description": "Leave unset for the template default, \"auto\" for a random combination that avoids your last one, \"brand\" for your saved colours, or an object of style axes.",
"oneOf": [
{
"type": "string",
"enum": [
"auto",
"brand"
]
},
{
"type": "object",
"properties": {
"palette": {
"type": "string"
},
"font": {
"type": "string"
},
"layout": {
"type": "string"
},
"shadow": {
"type": "boolean",
"description": "crisp offset shadow behind the type"
}
}
}
]
},
"scale": {
"type": "number",
"enum": [
1,
2,
3
],
"description": "device pixel ratio; 2 default, 3 costs $0.01 more"
},
"overlay": {
"oneOf": [
{
"type": "string",
"const": "default",
"description": "Apply your saved brand kit (set with PUT /v1/brand-kit)"
},
{
"type": "object",
"description": "Optional brand overlay composited into a corner after generation. Provide EITHER text OR logo_id (an uploaded brand logo), not both.",
"properties": {
"text": {
"type": "string",
"description": "Handle to render, ≤64 chars, single line (mutually exclusive with logo_id)"
},
"logo_id": {
"type": "string",
"description": "A logo uploaded via POST /v1/logos (mutually exclusive with text)"
},
"logo_style": {
"type": "string",
"enum": [
"sticker",
"shadow",
"outline",
"none"
],
"description": "Logo treatment (logo_id only). RECOMMENDED default \"sticker\": crisp die-cut border + drop shadow, legible on any photo. \"shadow\": drop shadow only (use when the logo already contrasts the scene). \"outline\": border only, flat cut-out. \"none\": bare logo, its own alpha only (use only for a pre-processed logo on a controlled background)."
},
"position": {
"type": "string",
"enum": [
"top-left",
"top-right",
"bottom-left",
"bottom-right"
],
"description": "default bottom-right"
},
"font": {
"type": "string",
"enum": [
"space-grotesk",
"inter",
"jetbrains-mono"
],
"description": "text only; default space-grotesk"
},
"weight": {
"type": "number",
"enum": [
400,
500,
700
],
"description": "text only; default 700"
},
"size": {
"type": "string",
"enum": [
"s",
"m",
"l"
],
"description": "text cap-height or logo width; default m"
},
"color": {
"type": "string",
"description": "text only; \"auto\" (default) or #RRGGBB"
},
"opacity": {
"type": "number",
"description": "0.2–1.0, default 0.9"
},
"margin": {
"description": "\"safe\" (default) or a non-negative px number"
}
}
}
]
}
},
"required": [
"template",
"slots"
]
}🟡upload_asset(data, kind)
Upload an image (PNG or JPEG, ≤5 MB) as a reusable asset — an avatar for a card, for example. Returns an ast_… id to pass into a template slot.
输入模式
{
"type": "object",
"properties": {
"data": {
"type": "string",
"description": "data URI or bare base64, e.g. \"data:image/png;base64,…\""
},
"kind": {
"type": "string",
"enum": [
"avatar",
"logo",
"photo",
"image"
],
"description": "what the image is for; \"avatar\" for a card byline"
}
},
"required": [
"data",
"kind"
]
}🟢list_models
List available image models with pricing and latency.
输入模式
{
"type": "object",
"properties": {}
}🟢list_templates
List card templates with their slots, style axes and exact price.
输入模式
{
"type": "object",
"properties": {}
}🟢get_balance
Show the current prepaid balance in USD.
输入模式
{
"type": "object",
"properties": {}
}🟢check_job(job_id)
Check an async generation job by its j_… id. Slow models (mode "async" in list_models) return a job instead of an image — poll this until it succeeds or fails.
输入模式
{
"type": "object",
"properties": {
"job_id": {
"type": "string",
"description": "j_… id returned by generate_image / edit_image"
}
},
"required": [
"job_id"
]
}社区
证据
最近观测
已验证未记录版本8 个工具
已验证未记录版本8 个工具