AGI-HDB : AI-Guided Intelligence for HDB Buying in Singapore.
An Agentic AI-powered property buying assistant and analysis system specifically designed for Singapore HDB (Housing Development Board) properties. It allows you to take a structured approach to your high-value purchase of an HDB flat in Singapore.
It allows you to brainstorm and evaluate properties using a 14-parameter framework but importantly it also allows you to consider and capture preferences and thoughts family members, short-term and long-term goals of family as you go through the buying process.
Prerequisites
This project uses Anthropic Claude APIs so you will need to have an Anthropic Claude account (they have a free tier which is good for testing).
Agentic AI Team Structure
The AI team consists of 6 specialized agents coordinated by a Lead Coordinator:
🎯 Lead Coordinator (Main Agent)
- Asks deep questions to understand the buyer’s requirements and preferences.
- Orchestrates the entire HDB buying process
- Maintains family members’ preferences and thoughts to guide throughout the buying process.
- Maintains a structured property shortlist in markdown format
- Makes final recommendations based on team analysis
- Coordinates between all AI team members
📊 HDB Market Research Analyst
- Singapore HDB market trends and price movements
- Ethnic quota analysis and availability tracking
- Neighborhood demographic analysis
- MRT proximity impact on property values
🏠 Property Evaluator
- Refers to purchaser’s family member preferences and thoughts.
- Individual HDB flat analysis using 14-parameter framework
- Comparative market analysis and COV calculations
- Property strengths and weaknesses assessment
- Value proposition evaluation for immediate occupation
💰 Financial Advisor
- HDB loan eligibility and calculations
- Bank mortgage options and interest rates
- CPF usage optimization strategies
- Total cash component and affordability assessments
🔨 Renovation & Extension Specialist
- HDB renovation guidelines and cost estimation
- Extension feasibility under HDB regulations
- HIP (Home Improvement Plan) year impact analysis
- Contractor recommendations and timeline assessment
📅 Viewing Coordinator
- Property viewing schedule management
- Viewing notes and photo documentation
- Agent coordination and follow-up tracking
- Property availability status monitoring
⚖️ Legal/HDB Documentation Specialist
- HDB resale procedures and legal requirements
- Contract review and legal documentation
- Regulatory compliance verification
- Legal risk assessment and buyer protection
Key Features
14-Parameter Evaluation Framework
Every property is analyzed using a comprehensive framework:
- Asking Price - Market positioning analysis
- Size (Sqft) - Space efficiency assessment
- Price per Sqft - Value comparison metrics
- Ethnic Quota - Availability and eligibility verification
- MRT Proximity - Distance and travel time analysis
- Number of Rooms - Layout and configuration evaluation
- COV (Cash Over Valuation) - Comparable transaction analysis
- Lease Decay - Remaining lease impact on value
- HIP Year - Home Improvement Plan timing considerations
- Total Cash Component - Upfront payment calculations
- Total Loan Component - Financing structure analysis
- Renovation Budget - Improvement cost estimation
- Extension Required - Structural modification needs
- Total Cash Out of Pocket - Complete financial commitment
Structured Property Tracking
- Maintains organized markdown-formatted property shortlist
- Real-time updates from all team members
- Comprehensive property status tracking
- Decision-making audit trail
Singapore HDB Expertise
- Specialized knowledge of HDB regulations and procedures
- Understanding of ethnic quota constraints
- MRT network impact analysis
- Government policy change lookup via search
Setup Instructions
Prerequisites
- Ruby 3.2.0 or higher
- Claude CLI installed and configured
- Bundler gem installed
Installation
- Clone or navigate to this project directory:
- Run the setup script:
This script will:
- Verify Ruby version (3.2.0+)
- Install Claude CLI if not present
- Install all dependencies with Bundler
- Verify configuration files
- Create initial property tracking files
Usage
Starting the Property Search Team
- Launch the Property Buying Team:
This will start the Lead Coordinator with connections to all specialist agents.
- Resume a previous session:
bin/launch --session-id [SESSION_ID]
- Launch with enhanced permissions:
- View active sessions:
Working with the Team
- Property Analysis Request: Ask the Lead Coordinator to analyze a specific property
- Market Research: Request neighborhood or market trend analysis
- Financial Assessment: Get detailed affordability and loan calculations
- Renovation Planning: Assess renovation needs and costs
- Viewing Management: Schedule and track property viewings
- Legal Review: Verify legal requirements and documentation
Property Shortlist Management
The Lead Coordinator maintains a structured markdown file (property-shortlist.md) that includes:
- All 14 evaluation parameters for each property
- Team assessment summaries
- Property status tracking
- Decision-making notes
Session Management
- All session data is stored in
~/.claude-swarm/sessions/
- Sessions can be resumed using
claude-swarm --session-id [SESSION_ID]
- View active sessions with
claude-swarm ps
Example Workflow
- Initialize Search: “I’m looking for a 4-room HDB flat in Tampines with good MRT access”
- Market Analysis: Team provides neighborhood analysis and price trends
- Property Evaluation: Individual properties are assessed using the 14-parameter framework
- Financial Planning: Complete financial analysis including loans and cash requirements
- Renovation Assessment: Renovation needs and costs are evaluated
- Decision Support: Comprehensive recommendation based on all factors
Contributing
This project is designed for Singapore HDB property search. Contributions welcome for:
- Enhanced property evaluation criteria
- Additional market analysis features
- Improved financial calculations
- Better renovation cost estimation
- MCPs for the major property search sites.
License
Refer to LICENSE.md for details.
Screenshots




