CodeTrace-AI

An AI-first development repository for building software through PRD-driven documentation, reusable system constraints, and staged specification generation. The project focuses on turning a single PRD into an executable document system that can later drive coding, testing, and acceptance.

Overview

CodeTrace-AI explores a modular workflow in which AI agents collaborate through explicit rules, skills, commands, memory, and staged specifications. The repository emphasizes:

• PRD-Driven Planning: Start from one PRD and derive the document system from actual project boundaries
• Specification-Driven Development: Build global, domain, testing, and acceptance documents before implementation
• Reusable AI Constraints: Keep system prompts, rules, commands, skills, and memory under version control
• Resumable Workflow: Track progress so work can continue from the latest completed document task

Documentation

• Product Requirements Document (PRD)
• AI Development Workflow
• PRD to Coding Orchestration
• PRD-Pipeline Interface
• Document System Todo Plan

Getting Started

This repository is designed for AI-assisted development where the document system comes first. The current setup focuses on planning and generating the specification tree from the PRD before moving into code.

Prerequisites

• Access to AI development tools that can read repository files and follow structured instructions
• A project PRD as the primary source of truth
• Basic familiarity with staged specification workflows

Usage

1. Read the PRD and Workflow.
2. Use docs/PRD.md, docs/Workflow.md, and .codex/modules/PRD-Pipeline/docs/ together to understand the intended document pipeline.
3. Use the reusable assets under .codex/modules/PRD-Pipeline/:
   • agents/ for system-prompt roles
   • skills/ for task-specific generation workflows
   • commands/ for reusable task entry points
   • rules/ for repository constraints
   • memory/ for stable project context
   • docs/ for internal orchestration and planning documents
4. Generate specs/ref/, specs/global/, specs/domains/, specs/testing/, and specs/acceptance/ in stages.
5. Use /prd-pipeline when you want the full document workflow to run end to end with checkpoint-based resume.

Project Structure

CodeTrace-AI/
├── .codex/
│   └── modules/
│       └── PRD-Pipeline/
│           ├── agents/
│           ├── commands/
│           ├── docs/
│           ├── memory/
│           ├── rules/
│           └── skills/
├── docs/
│   ├── PRD.md                          # Product Requirements Document
│   ├── Workflow.md                     # High-level workflow reference
├── LICENSE
└── README.md

Current Scope

The repository currently includes the document-system foundation for:

• PRD boundary analysis
• document-tree planning
• Phase 0 reference generation
• global specification generation
• domain planning and one-domain-at-a-time expansion
• testing-plan generation
• acceptance-criteria generation

Implementation code generation is intentionally downstream from these document phases.

Contributing

Contributions should preserve the AI-first workflow. Please:

1. Follow the rules in .codex/modules/PRD-Pipeline/rules/.
2. Keep reusable system-constraint files in English unless there is a documented exception.
3. Use a task-appropriate branch instead of reusing an unrelated branch.
4. Update progress-tracking documents when document-generation work is completed.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Inspired by modular AI-assisted development practices
• Built to demonstrate how PRD-driven document systems can structure later coding work