comfyui-mcp
通过MCP协议调用ComfyUI加载并执行工作流,支持文生图、图编辑、放大、去背景等场景。
README
comfyui-mcp
通过 MCP (Model Context Protocol) 协议调用 ComfyUI,加载并执行 workflow:文生图、图编辑、放大、去背景等。适用于 opencode、Claude Code、Cursor 等 MCP 客户端。
快速启动
uvx --from git+https://github.com/bookandmusic/comfyui-mcp.git comfyui-mcp \
--comfyui.host 127.0.0.1:8188
首次运行自动下载依赖并创建隔离虚拟环境。5 个内置 workflow 开箱即用,无需额外配置。
工具列表
内置 workflow(bundled_ 前缀)
| 工具名 | 说明 |
|---|---|
bundled_sensenova_json |
Sensenova 文生图 |
bundled_image_z_image_turbo_json |
Z-Image Turbo 文生图 |
bundled_qwen_image_2512_with_lora_json |
Qwen-Image-2512 + LoRA 文生图 |
bundled_utility_birefnet_remove_background_json |
BiRefNet 去背景 |
bundled_utility_z_image_turbo_2k_upscaler_json |
Z-Image Turbo 2K 放大 |
附加工具
| 工具名 | 说明 |
|---|---|
reboot_comfyui |
重启 ComfyUI 服务,等待恢复后返回成功消息 |
本地 workflow
放入 --comfyui.workflow_directory 的 .json 文件会自动注册为工具,工具名为文件名中的 . 替换为 _,例如 my_workflow.json → my_workflow。
各工具共有参数
| 参数 | 类型 | 说明 |
|---|---|---|
prompt |
string | 图像描述(非必填,由具体 workflow 决定) |
seed |
int | 随机种子,留空自动生成。多批时自动递增 |
batch_by_time |
bool | 按时间轮询(分钟),与 submit_batch 配合 |
submit_batch |
int | 重复执行次数,或 batch_by_time=true 时的轮询分钟数 |
output_dir |
string | 可选,本地目录。提供时将结果图片下载到此目录并返回 Markdown 图片链接 |
配置
comfyui-mcp 的所有配置通过命令行参数传入,不支持配置文件。启动参数如下:
启动参数
| 参数 | 默认值 | 说明 |
|---|---|---|
--comfyui.host |
$COMFYUI_HOST 或 127.0.0.1:8188 |
ComfyUI 服务的 host:port |
--comfyui.workflow_directory |
~/.config/comfyui-mcp/workflows |
本地工作流目录(sync 时也作为下载目标) |
--comfyui.additional_workflow_directories |
[] |
额外工作流目录列表 |
--comfyui.sync_from_comfyui |
false |
启动时从 ComfyUI 同步所有 .json 到本地 |
--comfyui.request_timeout |
30 |
HTTP 请求超时(秒) |
--comfyui.max_retries |
5 |
失败重试次数 |
--comfyui.queue_check_interval |
2.0 |
队列轮询间隔(秒) |
--comfyui.max_wait_time |
None |
任务最长等待时间(秒,None=无限) |
--comfyui.reboot_timeout |
300 |
reboot_comfyui 工具等待恢复时间(秒) |
Workflow 来源(按优先级)
1. 内置 workflow
随 wheel 打包,始终加载。工具名带 bundled_ 前缀,不与用户 workflow 冲突。
2. 本地目录
--comfyui.workflow_directory 指定的目录(默认 ~/.config/comfyui-mcp/workflows/)。放入 .json 文件,重启后自动注册。目录不存在时自动创建。
cp my-workflow.json ~/.config/comfyui-mcp/workflows/
需要多个目录时:
--comfyui.additional_workflow_directories '["/path/to/dir1", "/path/to/dir2"]'
3. 从 ComfyUI 同步(可选)
--comfyui.sync_from_comfyui true 启动后,MCP 调用 ComfyUI 的 /api/v2/userdata 列出所有 .json 文件并下载到本地 workflow_directory。适用于集中式 workflow 库 + 多 MCP 实例、跨机部署等场景。
uvx --from git+https://github.com/bookandmusic/comfyui-mcp.git comfyui-mcp \
--comfyui.host 127.0.0.1:8188 \
--comfyui.sync_from_comfyui true
注意:同步范围包括 userdata 下所有
.json(含非 workflow 的配置如styles.json),缺少nodes字段的文件会报加载错误但不中断启动。
客户端配置
opencode
{
"mcp": {
"comfyui": {
"type": "local",
"command": [
"uvx",
"--from", "git+https://github.com/bookandmusic/comfyui-mcp.git",
"comfyui-mcp",
"--comfyui.host", "127.0.0.1:8188"
],
"enabled": true,
"timeout": 600000
}
}
}
Claude Code
添加到 ~/.claude/settings.json:
{
"mcpServers": {
"comfyui": {
"command": "uvx",
"args": [
"--from", "git+https://github.com/bookandmusic/comfyui-mcp.git",
"comfyui-mcp",
"--comfyui.host", "127.0.0.1:8188"
]
}
}
}
也可放在项目级 .claude/settings.local.json 中。
本地开发
git clone https://github.com/bookandmusic/comfyui-mcp.git
cd comfyui-mcp
uv sync
uv run python -m comfyui_mcp.server --comfyui.host 127.0.0.1:8188
或安装为本地命令:
uv tool install .
comfyui-mcp --comfyui.host 127.0.0.1:8188
License
MIT — 详见 LICENSE。
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.