AI Agent Orchestrator

A SPARC-Based System for Intelligent Software Development

Project Overview

This project implements a sophisticated AI Agent Orchestration System based on the SPARC (Specification, Pseudocode, Architecture, Refinement, Completion) methodology. It leverages advanced Cognitive Triangulation for verification, a RULER-based quality evaluation system, and a suite of specialized domain agents to automate and manage complex software development workflows.

The core purpose is to coordinate multiple specialized AI agents to collaboratively design, implement, test, and validate software projects, ensuring alignment with user intent at every stage of development.

Key Features

  • SPARC Methodology: A structured five-phase workflow from specification to completion.
  • Cognitive Triangulation: Multi-stage verification to ensure stakeholder alignment and prevent requirement drift.
  • RULER Quality Evaluation: An LLM-as-Judge system to rank and select optimal implementation strategies.
  • Specialized Agents: 46 expert agents for tasks ranging from financial logic auditing to performance validation.
  • State Management: A robust SQLite-based memory system for comprehensive project history and context awareness.

Getting Started

To get started with the orchestrator, you need to have Python and Git installed. Clone the repository and install the required tools:

git clone https://github.com/jayfoeelite/custom-agents-orchestrator.git
cd custom-agents-orchestrator
pip install -r tools/requirements.txt

You can then begin interacting with the agent modes through a compatible AI coding assistant like RooCode.