sukoji

Sympo — AI skill for Claude Code

AI community

Multi-agent LLM orchestration for PRD-to-WBS generation with evaluation harness, ablations, and human study.

How to install Sympo

This entry records only its repository, not the path inside it, so there is no exact command to give. Open sukoji/sympo and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Sympo does

Multi-agent LLM orchestration for PRD-to-WBS generation with evaluation harness, ablations, and human study.

Alternatives in AI

ARIS Agent Guide — For AI agents reading this repo 6.2k ★
Open Multi Agent — TypeScript multi-agent orchestration engine — one runTeam() call from goal to result 5.8k ★
Orchestration Skills — Multi-agent orchestration patterns 5.3k ★

README

Multi-agent orchestration for intelligent work breakdown & assignment

*PRD, team context, and meeting notes in → debated WBS with roles, schedules, and buffers out.*

[![Python 3.10+](https://img.shields.io/badge/python-3.10+-3776AB?style=for-the-badge&logo=python&logoColor=white)](https://www.python.org/) [![LangGraph](https://img.shields.io/badge/LangGraph-orchestration-1f6feb?style=for-the-badge)](https://github.com/langchain-ai/langgraph) [![License: MIT](https://img.shields.io/badge/license-MIT-green?style=for-the-badge)](LICENSE)

**[한국어 README](README.ko.md)**

[Overview](#overview) · [Pipeline](#pipeline) · [Outputs](#outputs) · [Validation](#validation) · [Stack](#tech-stack) · [Run](#quick-start)

Overview

**symPO** (*symposium + project orchestration*) is a **research portfolio** exploring one question:

*Can LLM agents—given real planning inputs—negotiate their way to a **usable WBS**, not just a plausible one-shot draft?*

Instead of a single prompt, the system runs a **five-stage pipeline**: ingest context → draft structure → route specialists → debate adjustments → finalize assignments. A PM-style supervisor mediates buffers, reassignments, and convergence.

You provide	You receive
Product requirements	3-level work breakdown (phase → feature group → task)
Team skills & resumes	Role / responsibility mapping per leaf task
Meeting notes (optional)	Debate transcript showing how the plan evolved
Behavioral profiles (optional, experimental)	Quality scores (automatic + human-validated)

The repo ships the **full stack**: orchestration code, UI & API surfaces, MCP tool server, and an evaluation harness with ablation studies plus a **65-participant human survey**.