Project Butler - AI Agent for Project Documentation

Track

Creative Apps (GitHub Copilot)

Project Name

Project Butler

GitHub Username

huimingchen081-beep

Repository URL

https://github.com/huimingchen081-beep/project-butler

Project Description

Project Butler - AI Agent for Project Documentation

Project Butler is an AI-powered agent that generates complete project documentation from natural language descriptions. Tell it your project idea, and it produces PRDs, competitive analysis, financial models, pitch decks, and technical architecture documents — all in minutes.

Key Features

• 📋 PRD Generation: Product Requirements Documents with user stories, acceptance criteria, and feature specs
• 📊 Competitive Analysis: Market landscape, competitor comparison, SWOT analysis
• 💰 Financial Models: Revenue projections, cost structures, unit economics (Excel output)
• 🎯 Pitch Decks: Investor-ready presentation slides (PPTX output)
• 🏗️ Tech Architecture: System design, tech stack recommendations, API specifications

How It Works

1. User inputs a project description in natural language
2. AI Planning Engine breaks down requirements into structured tasks
3. Sequential pipeline executes 7 specialized agents:
   • Research Agent → Outline Agent → PRD Agent → Competitive Agent → Financial Agent → Pitch Agent → Architecture Agent
4. Each agent validates outputs before passing to the next stage
5. Final deliverables: PPTX, DOCX, XLSX files ready for download

Tech Stack

• Runtime: Node.js (CLI tool)
• LLM: DashScope / Qwen (Alibaba Cloud) via compatible-mode API
• Output: python-pptx, python-docx, openpyxl for real Office file generation
• Packaging: Node.js archiver for ZIP bundle of all deliverables

Demo Video

https://youtu.be/KXQiskRYGeM

Repository

https://github.com/huimingchen081-beep/project-butler

Demo Video or Screenshots

Demo Video: https://youtu.be/KXQiskRYGeM Screenshots: https://github.com/huimingchen081-beep/project-butler Live Demo: N/A (CLI tool)

Primary Programming Language

TypeScript/JavaScript

Key Technologies Used

• Node.js (runtime, CLI tool)
• DashScope API / Qwen LLM (Alibaba Cloud AI)
• python-pptx / python-docx / openpyxl (Office document generation)
• archiver (ZIP packaging)

Submission Type

Individual

Team Members

No response

Submission Requirements

• My project meets the track-specific challenge requirements
• My repository includes a comprehensive README.md with setup instructions
• My code does not contain hardcoded API keys or secrets
• I have included demo materials (video or screenshots)
• My project is my own work with proper attribution for any third-party code
• I agree to the Code of Conduct
• I have read and agree to the Disclaimer
• My submission does NOT contain any confidential, proprietary, or sensitive information
• I confirm I have the rights to submit this content and grant the necessary licenses

Quick Setup Summary

Clone repo: git clone https://github.com/huimingchen081-beep/project-butler.git
Install dependencies: npm install && pip install python-pptx python-docx openpyxl
Configure env: copy .env.example to .env, set DASHSCOPE_API_KEY
Run: node agent.js "your project description"
Output files in output/ directory (ZIP bundle with PPTX, DOCX, XLSX)

Technical Highlights

Designed a 7-stage sequential Agent pipeline (Research > Outline > PRD > Competitive > Financial > Pitch > Architecture), with each agent running independently, validating output before passing to the next stage
Built an LLM-powered intelligent task planning engine that decomposes natural language into structured document generation tasks
Generates real editable Office files using Python libraries (python-pptx/python-docx/openpyxl), not fake PDF wrappers
API keys managed via .env — zero hardcoded secrets
Single CLI entry point, zero-config usage

Challenges & Learnings

The biggest challenge was ensuring output quality and format consistency across 7 collaborating agents. Each agent's output feeds into the next — a single error cascades through the entire pipeline. The solution was embedding self-validation into every agent, verifying output format and completeness before proceeding to the next stage.

Another challenge was Office file generation. Pure JavaScript solutions proved unreliable, so we adopted a Node.js + Python hybrid architecture: Node.js handles LLM orchestration and flow control, while Python handles professional document format generation — each doing what it does best.

Contact Information

chmchm811@163.com

Country/Region

China

Start with the linked Project Butler repository and its README.md, then inspect the agent.js CLI entry point and the Node.js/Python setup described in the issue. Run the documented npm and pip installation steps before trying the sample command. Done means the seven-stage pipeline produces the stated editable Office files and ZIP output without exposed API keys.