Serein Flow
可扩展的节点流编辑器与工作流引擎,内置 MCP,支持 AI 自动生成节点与流程。An extensible node-flow editor and workflow engine with built-in MCP, enabling AI-powered generation of nodes and workfl…
- Transport
- Not stated
- Package
- —
- Registry id
- —
No install snippet on purpose. A working MCP config is a command, its arguments and an environment block — the last two are where API keys live, so this catalogue never stores them and cannot publish them. Follow the link above for the authors' own instructions.
Documentation portal: SereinFlow tutorials and current API reference (Chinese).
SereinFlow is a web-first workflow orchestration and execution system. It combines a visual web console, an ASP.NET Core API, isolated Worker Runner processes, and an MCP service for AI-tool integration.
Project Overview
SereinFlow is used to create, maintain, and run project workflows. The web console manages projects, flows, versions, runs, and environment settings. The API coordinates persisted data and execution requests, while untrusted scripts and plugins run only in separate Worker Runner processes. The MCP service exposes controlled tools, resources, and prompts so AI clients can work with SereinFlow safely.
| Path | Purpose |
|---|---|
src/SereinFlow.Api |
ASP.NET Core API and the HTTP/stdio MCP host. |
src/SereinFlow.SDK |
Typed HTTP client SDK for third-party applications. |
frontend/sereinflow-web |
Vue + Vite web console. |
plugins/sereinflow-ai-toolkit |
Local MCP client and Skill source material. |
docs |
Maintained technical references. |
Community & Technical Support
If you need help using SereinFlow, join the QQ group 955830545 to contact me for technical support, usage discussions, or issue feedback.
Prerequisites
- .NET SDK
10.0.100or a compatible SDK selected byglobal.json. - Node.js
24.19.0, as defined in.node-version, and npm11.6.2. - Windows is assumed for the environment-variable and Codex instructions below. The API and frontend commands are otherwise standard .NET and npm commands.
Projects that use SereinLang rely on the pinned source snapshot under
src/ThirdParty/SereinScript. A normal SereinFlow build needs neither a
separate SereinScript checkout nor a path environment variable.
Run Locally
1. Start the API
From the repository root, run:
dotnet run --project .\src\SereinFlow.Api\SereinFlow.Api.csproj
By default, the API and HTTP MCP service listen on http://127.0.0.1:8188,
with MCP available at http://127.0.0.1:8188/mcp. The first start initializes
local data under the configured data directory.
2. Start the Web Console
Open a second terminal and run:
Set-Location .\frontend\sereinflow-web
npm ci
npm run dev
Open the URL printed by Vite, usually http://localhost:5173. The development
server proxies /api and /hubs to http://127.0.0.1:8188 by default. Set
VITE_API_PROXY_TARGET before starting the frontend when the API uses another
address.
Configure the MCP Service
The following workflow configures an HTTP MCP client for a locally running SereinFlow instance. Start both the API and the web console before continuing.
1. Generate the Administrator Key
In the web console's home menu, open Environment settings, find MCP
access keys, and select Generate first key. Save the administrator key
shown there. It normally begins with sfk_..., and the complete Secret is
shown only once when it is created or rotated.
2. Add the Environment Variable
In Windows Edit environment variables for your account, add a user variable
named SEREINFLOW_MCP_API_KEY whose value is the complete key you just
generated. Alternatively, run this PowerShell command, replacing the example
value with the real key:
setx SEREINFLOW_MCP_API_KEY "sfk_replace_with_your_complete_secret"
After running setx or saving the system setting, close and reopen terminals
and AI tools. Existing processes do not receive the new variable automatically.
3. Create a Local Skill for Your AI Tool
From the project's README.