An Event-Driven AI Coding Assistant Core Engine

Providing reliable and pluggable intelligent processing capabilities for building code assistant tools.

中文 | English

📖 Project Overview

Sema Core is an event-driven AI coding assistant core engine that provides reliable and pluggable intelligent processing capabilities for building code assistant tools. It supports core capabilities such as multi-agent collaboration, Skill extension, and Plan mode task planning, and can be quickly integrated into various AI programming tools.

View Documentation

💼 Use Cases

SemaCode Extension (VSCode / JetBrains)

SemaCode Extension is an intelligent programming plugin based on the Sema Core engine, supporting both the VSCode extension and JetBrains plugins (IntelliJ IDEA / PyCharm / GoLand / WebStorm / CLion, etc.).

SemaWork

SemaWork is a local web UI built on Sema Core.

SemaPLC

SemaPLC is an IDE for industrial PLC programming — describe a control task in natural language and watch it become a running PLC program, powered by the embedded Sema Core Agent.

SemaClaw

SemaClaw is a general-purpose engineering harness for building personal AI agents.

🚀 Quick Start

Get SemaWork running locally in two steps:

1. Clone and build

git clone https://github.com/midea-ai/sema-code-core.git
cd sema-code-core && npm install && npm run build

2. Start SemaWork

cd webui && npm install && npm run build
npm start

Then open the URL printed in the terminal (http://127.0.0.1:3210/?token=…) in your browser, or run npm start -- --open to open it automatically.

🛠 Development

# 1. Install dependencies
npm install

# 2. Build
npm run build

# 3. Pack (optional): creates sema-core-x.y.z.tgz for local testing via npm install <tgz>
npm pack

Non-Node projects can integrate via the multi-language SDKs (Java / Python / C#).

🌐 The Sema Family

Midea AIRC · SEMA Agent Systems explores agent systems engineering through three open-source research efforts. Sema Code turns product-bound coding agents into programmable, embeddable infrastructure; SemaClaw studies harness engineering for open, controllable, and extensible personal agents; and SemaPLC grounds industrial code generation in existing projects and gates completion on specification, compilation, and runtime evidence. Embed · Harness · Verify summarizes the distinct focus of each project, not a shared pipeline or strict dependency chain.

The SEMA series currently includes the following papers and open-source implementations:

[1] Sema Code: Decoupling AI Coding Agents into Programmable, Embeddable Infrastructure · Code
[2] SemaClaw: A Step Towards General-Purpose Personal AI Agents through Harness Engineering · Code
[3] SemaPLC: A Project-Grounded, Verification-Gated Agent Harness for PLC Code Generation · Code

📜 License

sema-core is released under the MIT License. See THIRD_PARTY_NOTICES.md for the third-party dependency list and external service terms.