Skip to content

dtehan/AgentBuilder

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AgentBuilder

A framework for building specialized AI agents that work with the Teradata platform. AgentBuilder provides templates, configurations, and agent definitions for developing domain-specific agents powered by Claude Code.

Overview

AgentBuilder enables the creation of intelligent agents that can automate complex Teradata operations including database administration, security auditing, space management, SQL optimization, and more. The framework leverages Claude Code's sub-agent architecture and integrates with the teradataMCP server for comprehensive database operations.

Project Structure

AgentBuilder/
├── .claude/
│   ├── agents/              # Agent definitions (sub-agents)
│   │   ├── teradata/        # Teradata-specific agents
│   │   │   ├── teradata-code-generator.md           # Main routing agent (entry point)
│   │   │   ├── teradata-security-auditor.md
│   │   │   ├── teradata-space-manager.md
│   │   │   ├── compression-advisor.md
│   │   │   └── teradata-statistics-collector.md
│   │   ├── business/        # Business domain agents
│   │   │   └── retail-analytics.md
│   │   └── util/            # Utility agents
│   │       ├── meta-agent.md          # Main agent for building new agents
│   │       ├── load_doc_agent.md      # Agent for loading documentation from URLs
│   │       ├── build_schemas.md       # Agent for loading database schema information
│   │       └── build_DBHierarchy.md   # Agent for generating database hierarchy docs
│   ├── commands/            # Slash commands
│   │   └── utils/
│   │       └── all_skills.md         # Lists all available skills
│   └── skills/              # Project skills (model-invoked capabilities)
│       ├── fork-terminal/            # Terminal forking skill
│       │   ├── SKILL.md              # Skill definition
│       │   ├── tools/
│       │   │   └── fork_terminal.py  # Terminal forking utility
│       │   ├── prompts/              # Prompt templates
│       │   └── cookbooks/            # Usage examples
│       └── another_skill/
│           └── SKILL.md
├── app/                     # TeradataCodeGenerator prompt library
│   └── TeradataCodeGenerator/
│       ├── FunctionalPrompts/
│       │   ├── Core_SQL_Functions/      # 126+ SQL function prompts
│       │   ├── Advanced_Analytics/      # Advanced analytics prompts
│       │   └── INDEX.md                 # Function index
│       └── ProcessPrompts/
│           ├── persona_dba.md           # DBA persona routing
│           ├── persona_data_scientist.md # Data Scientist persona routing
│           ├── persona_data_engineer.md  # Data Engineer persona routing
│           ├── dba/                     # DBA process workflows (6)
│           └── ml/                      # ML process workflows
├── ai_docs/                 # Documentation resources
│   ├── README.md            # Document source index
│   ├── doc_overview.md
│   ├── doc_sub-agents.md
│   ├── doc_skills.md
│   ├── doc_output-styles.md
│   ├── doc_hooks-guide.md
│   ├── doc_headless.md
│   ├── doc_mcp.md
│   ├── doc_teradataMCP.md
│   └── doc_TDSQL.md
├── .mcp.json               # MCP server configuration
├── AGENTS.md               # Agent documentation
├── CLAUDE.md               # Project instructions for AI
├── settings.json           # Permission settings
└── README.md               # This file

Available Agents

Teradata-Specific Agents

Agent: teradata-security-auditor

Color: Red | Model: Sonnet

A specialist agent for generating comprehensive security reports and analyzing user permissions within Teradata environments.

Capabilities:

  • Analyzes database permissions and role assignments
  • Reviews role hierarchies and inherited permissions
  • Identifies security risks (excessive permissions, privilege escalation, grant option abuse)
  • Generates actionable recommendations with SQL remediation scripts
  • Performs comparative multi-user security analysis
  • Provides risk categorization (Critical, High, Medium, Low)

Use Cases:

  • Security compliance audits
  • User access reviews
  • Permission cleanup and optimization
  • Identifying separation of duties violations
  • Security hardening initiatives

Agent: teradata-space-manager

Color: Orange | Model: Sonnet

A DBA specialist agent focused on database space management and optimization.

Capabilities:

  • Monitors space utilization across all databases
  • Identifies databases at risk of running out of space
  • Generates space reallocation recommendations
  • Creates SQL scripts for space modifications
  • Analyzes database hierarchy for space reallocation
  • Categorizes databases by risk level (Critical >90%, Warning 80-90%, Healthy <80%)

Use Cases:

  • Proactive space management
  • Preventing out-of-space errors
  • Space allocation optimization
  • Capacity planning
  • Database cleanup identification

Agent: compression-advisor

Model: Default

An analytics specialist agent that performs rigorous cost-benefit analysis to identify optimal Multi-Value Compression (MVC) candidates.

Capabilities:

  • Analyzes tables using compression equation for accurate space savings
  • Identifies high-cardinality columns suitable for compression
  • Generates production-ready ALTER TABLE statements
  • Calculates compression ratios and storage optimization
  • Provides cost-benefit analysis for compression decisions

Use Cases:

  • Storage optimization initiatives
  • Reducing table space consumption
  • Identifying compression opportunities across databases
  • Performance improvement through I/O reduction
  • Cost reduction in cloud environments

Agent: teradata-statistics-collector

Model: Default

A specialist agent for Teradata statistics analysis and maintenance.

Capabilities:

  • Identifies missing or stale statistics
  • Analyzes statistics health across databases
  • Generates optimized COLLECT STATS recommendations
  • Improves query optimizer performance
  • Provides statistics maintenance schedules

Use Cases:

  • Query performance optimization
  • Statistics health assessment
  • Proactive statistics maintenance
  • Troubleshooting slow queries
  • Database performance tuning

Agent: teradata-code-generator

Color: Orange | Model: Sonnet

An intelligent routing agent that serves as the main entry point for all Teradata assistance. Uses progressive prompting to route requests to specialized personas based on task analysis.

Capabilities:

  • Routes requests to appropriate persona (DBA, Data Scientist, Data Engineer)
  • Analyzes user intent from keywords and context
  • Progressive prompting through hierarchical workflow structure
  • Access to 126+ documented Teradata SQL functions
  • Coordinates multi-persona workflows for complex tasks
  • Direct SQL function lookup and documentation

Personas:

  1. Database Administrator (DBA): Health assessments, performance monitoring, lineage, data quality, archiving, metadata, maintenance
  2. Data Scientist: ML model development, statistical analysis, predictive modeling, feature engineering, model evaluation
  3. Data Engineer: Data preparation, ETL/ELT pipelines, data transformation, quality validation, missing value handling, outlier detection

Use Cases:

  • Unified entry point for all Teradata tasks
  • Automatic routing to domain experts
  • Complex workflows requiring multiple specializations
  • SQL function documentation and examples
  • End-to-end data science pipelines (prep → train → deploy)

Business Domain Agents

Agent: retail-analytics

Model: Default

A business intelligence specialist for analyzing retail data and generating insights.

Capabilities:

  • Analyzes sales, customer, product, and inventory data
  • Generates business reports and dashboards
  • Performs statistical analysis on retail datasets
  • Answers questions about retail_sample_data schema
  • Creates data-driven business insights

Use Cases:

  • Sales performance analysis
  • Customer behavior insights
  • Product performance tracking
  • Inventory optimization
  • Business intelligence reporting

Utility Agents

Agent: meta-agent

Color: Cyan | Model: Opus

An expert agent architect that generates new sub-agent configuration files from user descriptions.

Capabilities:

  • Analyzes requirements and generates complete agent definitions
  • Selects appropriate tools and models for new agents
  • Creates structured markdown agent configuration files
  • Incorporates best practices and proper delegation descriptions
  • Writes agents to the appropriate directory structure

Use Cases:

  • Rapid agent prototyping
  • Creating domain-specific agents
  • Extending the framework with new capabilities

Agent: load_doc_agent

Model: Default

A research specialist for fetching and managing documentation resources.

Capabilities:

  • Fetches markdown documentation from specified URLs
  • Checks for recently updated documents (30-day threshold)
  • Supports parallel document fetching
  • Stores structured documentation in ai_docs/ directory
  • Integrates with Firecrawl for web scraping

Use Cases:

  • Keeping documentation up to date
  • Loading AI Docs resources
  • Managing knowledge base content

Agent: build_schemas

Model: Default

A database schema documentation generator that creates comprehensive markdown documentation from existing SQL database schemas.

Capabilities:

  • Generates structured markdown documentation for database schemas
  • Extracts table structures, column descriptions, and relationships
  • Documents table usage patterns and affinities
  • Creates preview data samples
  • Integrates with teradataMCP for schema analysis

Use Cases:

  • Building business agents that require database schema knowledge
  • Database documentation generation
  • Onboarding new developers to data models
  • Data governance and lineage tracking
  • Schema change management

Agent: build_DBHierarchy

Model: Default

A database hierarchy documentation generator that creates structured documentation of database relationships.

Capabilities:

  • Generates database hierarchy documentation in markdown format
  • Maps parent-child relationships between databases
  • Documents database structure and organization
  • Creates visual hierarchy representations
  • Integrates with teradataMCP for metadata extraction

Use Cases:

  • Understanding database architecture
  • Database organization documentation
  • Impact analysis for database changes
  • Security boundary identification
  • Data governance hierarchy mapping

Available Skills/Commands

Skills are markdown files that extend Claude's capabilities with specialized knowledge for specific tasks. They are model-invoked, meaning Claude automatically decides which skills to use based on your request.

For complete information about creating and using skills, see doc_skills.md.

Current Skills

/utils:all_skills

Lists all available skills loaded in your system prompt.

Usage:

/utils:all_skills

fork-terminal

Creates a forked terminal session within Claude Code or CLI to run tasks.

Usage: Use when you need to:

  • Create a new terminal session
  • Fork an existing session
  • Run commands in an isolated terminal environment

Trigger phrases:

fork a terminal
create a terminal
fork session
new terminal: <command>

Creating New Skills

Skills are stored in:

  • Project Skills: .claude/commands/ (shared with repository)
  • Personal Skills: ~/.claude/skills/ (available across all projects)

To create a new skill:

  1. Create a directory in .claude/commands/ or ~/.claude/skills/
  2. Add a SKILL.md file with YAML metadata and instructions
  3. Restart Claude Code to load the new skill

Example structure:

---
name: my-skill-name
description: Brief description of what this skill does and when to use it
---

# Instructions
Provide clear guidance for Claude on how to use this skill.

For detailed skill authoring guidance, see doc_skills.md.

TeradataCodeGenerator Prompt Library

The app/TeradataCodeGenerator/ directory contains a comprehensive library of reusable prompts and templates for Teradata operations, organized using a progressive prompting architecture.

Progressive Prompting Architecture

The TeradataCodeGenerator uses a hierarchical routing system:

Entry Point → Persona Files → Process Prompts → Functional Prompts

Flow:

  1. teradata-code-generator.md (Entry Point) - Analyzes user intent and routes to appropriate persona
  2. Persona Files - Specialized experts (DBA, Data Scientist, Data Engineer)
  3. Process Prompts - Step-by-step workflows for specific tasks
  4. Functional Prompts - Detailed documentation for individual SQL functions (126+ functions)

Personas

Database Administrator (DBA) (ProcessPrompts/persona_dba.md)

  • Health assessments, performance monitoring, lineage analysis
  • Data quality assessments, archiving strategies
  • Business metadata and documentation
  • Database maintenance and optimization
  • 6 DBA process workflows available

Data Scientist (ProcessPrompts/persona_data_scientist.md)

  • Machine learning model development and training
  • Statistical analysis and hypothesis testing
  • Predictive modeling, feature engineering
  • Model evaluation and performance metrics
  • ML process workflows expanding

Data Engineer (ProcessPrompts/persona_data_engineer.md)

  • Data preparation and ETL/ELT pipelines
  • Data transformation and quality validation
  • Missing value handling and outlier detection
  • Data profiling and integration
  • Data engineering workflows

Process Prompts

DBA Process Prompts (ProcessPrompts/dba/) Process-oriented prompts for database administration tasks including maintenance, monitoring, and optimization workflows.

ML Process Prompts (ProcessPrompts/ml/) Machine learning and predictive analytics workflow templates for Teradata ML capabilities.

Functional Prompts (126+ Functions)

Core SQL Functions (40+ prompts) Detailed prompt templates for Teradata SQL functions including:

  • Date/time functions (current_date, add_months, extract)
  • Aggregate functions (avg, sum, count, max, min)
  • Statistical functions (stddev_samp, var_pop, kurtosis)
  • Window functions (rank, lead, percent_rank)
  • String and mathematical operations (mod, ceil, nvl, coalesce)
  • Data transformation (pivot, decode, greatest, least)

Advanced Analytics Templates for complex analytical operations and advanced SQL patterns.

Usage

These prompts serve as building blocks for:

  • Intelligent routing to domain-specific expertise
  • Progressive workflow execution (entry → persona → process → function)
  • Multi-persona task coordination
  • Standardizing SQL function usage with 126+ documented functions
  • Ensuring best practices in Teradata SQL
  • End-to-end data science pipelines
  • Training and documentation purposes

MCP Integration

teradata-code-generator connects to a teradataMCP server providing comprehensive tooling for Teradata operations:

For complete list of MCP capabilities, refer to the teradataMCP documentation.

Configuration

MCP Server Connection

{
  "teradataMCP": {
    "type": "http",
    "url": "http://MCP_host_address:8001/mcp/",
    "env": {
      "DATABASE_URI": "teradata://username:password@Database_host_address:1025/default_database"
    }
  }
}

Getting Started

  1. Ensure you have Claude Code installed and configured
  2. Clone this repository
  3. Configure your Teradata connection in .mcp.json
  4. Invoke agents using the @agent-name syntax in Claude Code
  5. Use skills with /skill-name syntax

Usage Examples

teradata-code-generator (Unified Entry Point)

@teradata-code-generator Check the health of my production database
@teradata-code-generator How do I train an XGBoost model for customer churn prediction?
@teradata-code-generator I need to clean my dataset and handle missing values before analysis
@teradata-code-generator How do I calculate moving averages in Teradata?

Security Audit

@teradata-security-auditor analyze security for user john_smith

Space Management

@teradata-space-manager check database space utilization and identify at-risk databases

Compression Analysis

@compression-advisor analyze compression opportunities for database retail_db

Statistics Health Check

@teradata-statistics-collector analyze statistics for database analytics_db

Retail Analytics

@retail-analytics analyze top 10 products by revenue for Q4 2024

Build Database Schema Documentation

@build_schemas generate schema documentation for database customer_data

Build Database Hierarchy

@build_DBHierarchy document the hierarchy for database prod_environment

Load Documentation

/utils:load_ai_docs

Extending the Framework

To create a new agent:

  1. Use the meta-agent to generate a template: @meta-agent create an agent that [description]
  2. Review and customize the generated agent definition
  3. Place the agent file in the appropriate directory (.claude/agents/teradata/ or .claude/agents/util/)
  4. Update agent metadata (name, description, tools, model, color)
  5. Test the agent with sample scenarios

Create New Agent

@meta-agent create an agent that monitors query performance and sends alerts

About

Agent Builder Repo

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages