agent-artifex

Use when the user asks "what testing do we need?", "what are our testing gaps?", "we have some tests but are they enough?", "is our MCP server well-tested?", "what should we test next?", "audit our test coverage for AI", "we keep getting bad responses and don't know why", "our agent picks the wrong tool sometimes", "how is my tool design?", "are my descriptions good enough?", "review my error messages", "is my system prompt well-designed?", "audit my MCP server design", "what design gaps do I have?", or needs to diagnose AI design or testing gaps in an existing project. Also use when someone says "assess my testing", "review our test strategy for AI", or "assess my design".

Use when the user wants to design an MCP server, agent, chatbot, or tool-calling system for quality. This includes: designing tool descriptions, structuring parameters and schemas, writing error messages for LLM consumers, designing system prompts, planning multi-turn conversations, architecting tool sets, or designing response formats. Also use when someone says "how should I design", "what makes a good tool description", "how should I structure my errors", "design my MCP server", "how do I organize my tools", or any task where they want to follow evidence-based design principles before or while building.

Use when the user asks "what is the AI testing framework?", "what are the design principles?", "explain the 7 design areas", "explain MCP testing", "what are the testing areas?", "what's the testing pyramid for AI?", "how do the testing layers relate?", "what should I test in my MCP server?", "overview of AI agent design and testing", or needs a comprehensive reference overview of the AI design and testing guidelines. Also use when someone is new to designing or testing AI systems and wants the big picture before diving into implementation.

Use when the user is unsure which AI services skill to invoke, asks for "AI services guidance" generally, says "help me design my MCP server", "how should I structure my tools?", "what makes a good tool description?", "how do I design my agent?", "help me test my MCP server", "how should I test my agent?", "what testing do I need?", "my tests are flaky", "the agent picks the wrong tool", "how do I set up CI for AI tests?", or wants a structured guided experience that routes across multiple AI services skills. Also use when the user mentions designing or testing chatbots, tool descriptions, evals, or agent behavior without specifying a particular skill.

Use when the user wants to improve an existing MCP server, agent, chatbot, or tool-calling system. This includes: improving tool descriptions, fixing error messages, adding output schemas, writing tests, implementing quality checks, adding evals, setting up test harnesses, or any task where they say "help me improve", "fix my descriptions", "add tests", "write evals", "implement quality checks", "make my server better", "apply the design principles", or are ready to make code changes to improve quality. This skill covers both design application (making the code better) and test implementation (verifying the code is good). For scaffolding new projects, use claude-api:mcp-builder. For design principles without code changes, use agent-artifex:design.

Use when the user says "I'm new to AI testing", "teach me about designing MCP servers", "how do I design good tool descriptions?", "walk me through design principles", "explain the design areas", "explain the testing pyramid for agents", "how do I test tool descriptions?", "walk me through an example", "I read the docs but it's not clicking", "how do evals work?", "what's faithfulness in AI testing?", "explain the causal chain", or wants to build fluency in AI services design and testing through Socratic dialogue rather than reading reference material.