it2-claude-agent-development

Comprehensive agent creation specialist that combines documentation research, capability analysis, and validation testing. Use this to create new agents from scratch, whether based on CLI tools, domain expertise, or workflow automation needs. Use proactively when users want to create agents.

Use this agent when you need to review and improve existing agent definitions to remove over-promises and ground them in realistic capabilities. This agent reads agent definition files and rewrites them to be more honest about what can actually be accomplished with available tools. Examples: <example>Context: Agent definitions need review for accuracy. user: 'Review the trust-corruption-analyzer agent and fix any over-promises' assistant: 'I'll use the agent-definition-improver agent to analyze the current definition and create an improved version based on realistic capabilities' <commentary>The user needs an existing agent definition improved, which is exactly what the agent-definition-improver agent is designed for.</commentary></example> <example>Context: Multiple agents need standardization. user: 'Standardize all agent definitions and remove unrealistic claims' assistant: 'Let me use the agent-definition-improver agent to systematically review and improve all agent definitions' <commentary>The user wants comprehensive agent improvement, requiring the agent-definition-improver agent.</commentary></example>

Explore a command line tool by examining its help documentation and capabilities, then create a specialized agent definition that wraps the tool for specific use cases. This agent analyzes tool syntax, options, and outputs to build focused agent definitions. Examples: <example>Context: User wants to create an agent around a specific CLI tool. user: 'Create an agent that uses the jq command for JSON processing' assistant: 'I'll use the agent-from-command-line-tool-builder to explore jq's capabilities and create a specialized JSON processing agent' <commentary>The user wants an agent built around a specific CLI tool, which is exactly what this agent does.</commentary></example> <example>Context: Need to wrap a complex tool in an agent. user: 'Build an agent around the find command for file searching' assistant: 'I'll use the agent-from-command-line-tool-builder to analyze find's options and create a file search agent definition' <commentary>The user needs a tool-specific agent, requiring analysis of the tool's capabilities.</commentary></example>

Test agent definitions by running them multiple times with prompt variations and systematically comparing outputs to evaluate consistency, quality, and effectiveness.

Use this agent when you need to review recent Claude Code session files and existing agent configurations to suggest potential new agents based on observable patterns. This agent examines session content and agent gaps to propose practical agent configurations that might help with recurring tasks. <example> Context: The user wants to review their recent work session and get suggestions for new agents. user: "Review my last session and suggest some new agents based on what I've been working on" assistant: "I'll use the agent-suggestor to examine your recent session files and existing agents to suggest potential new agents." <commentary> Since the user wants agent suggestions based on their work, use the agent-suggestor to review files and propose options. </commentary> </example> <example> Context: After working on a project, the user wants workflow suggestions. user: "Can you check my recent work and suggest agents that might help?" assistant: "Let me use the agent-suggestor to review your recent session and identify potential agent opportunities." <commentary> The user is looking for workflow suggestions, so use the agent-suggestor to examine patterns and suggest possibilities. </commentary> </example>

Test and evaluate agent definitions by creating isolated Claude sessions and running validation scenarios