lyxhnu

lyxhnu / multi-coding-agent

Public

一个“全透明的多 Agent Coding 系统”，它不是传统那种单个大模型假装多个角色协作，而是把一个完整的软件交付过程拆成了真实的多个 Agent 来做。 我把需求分析、架构设计、前端开发、后端开发、环境与部署、测试验证分别交给 PC、CA、FD、BD、DE、QT 六类角色，每个角色都有独立身份、独立会话、独立状态和独立任务。

18
0
100% credibility
Found Mar 21, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A web interface for a multi-agent AI system that builds software projects from natural language requests by coordinating specialized agents to plan, code, test, and deliver with real-time visibility.

How It Works

1
🔍 Discover the coding helper

You find this friendly web app that turns your ideas into working code using a team of smart helpers.

2
💭 Describe your project

You type a simple description of what you want to build, like a game or a web tool, in the chat box.

3
🚀 Launch the team

Hit the button and watch the team of helpers spring into action, each taking on their special role to build your project.

4
👀 Follow the action live

See every step unfold in real time: thoughts, teamwork, changes, and progress updates right on screen.

5
📁 Check and tweak files

Peek inside the growing project files, make quick edits if needed, and save your changes easily.

🎉 Your project is ready

Celebrate as your fully built app comes to life, complete and working just as you imagined.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is multi-coding-agent?

This Python-based agent coding AI spins up a full software delivery pipeline from a single natural language request, assigning tasks to six specialized agents: product coordinator, architect, frontend dev, backend dev, deploy engineer, and QA tester. Each agent runs in isolation with its own session, tools for file I/O, terminal, REPL, and inter-agent messaging, building a sandboxed Next.js + FastAPI project while streaming progress via a web UI. It solves the "fake multi-agent" problem in tools like Cursor multi agent coding by enforcing real role separation and task handoffs.

Why is it gaining traction?

Stands out from agent GitHub Copilot or agent GitHub Claude clones with true parallelism—multiple agents work concurrently on design, code, deploy, and test—plus durable state for resuming interrupted runs. Devs dig the transparent task board, live agent chats, and artifact generation without IDE plugins; it's a free agent coding GitHub repo benchmarking real-world agent coding protocol against single-model hacks. Reddit threads on agent coding tips praise its web search/RAG skills and LLM flexibility (OpenAI, DeepSeek, etc.).

Who should use this?

Fullstack prototyper wanting agent coding free for MVPs, like "build a game dashboard with API backend." AI researchers benchmarking agent coding AI on end-to-end workflows. Indie devs bypassing boilerplate in agent GitHub action setups or Cursor 2.0 multi agent coding experiments.

Verdict

Promising early experiment (18 stars, 1.0% credibility) for agent coding reddit fans, but skip for production—docs are thin, no tests visible. Try for weekend hacks if you configure your LLM keys.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.