What problem does it solve? Designing well-structured AI agent skills is hard: agents often end up over-specified with mechanical procedures, lack clear identity, or have no coherent memory architecture. This Skill guides users through a structured conversational process to create outcome-driven agents with clear personas, focused capabilities, and optional persistent memory. ## Core Features & Use Cases - Agent Building: A six-phase discovery process (intent, capabilities strategy, requirements, drafting, building, summary) that produces stateless, memory, or autonomous agent types with sanctum templates, First Breath onboarding, and optional PULSE autonomous behavior. - Quality Analysis: Analyzes existing agents for over-specification, persona-capability misalignment, and structural issues, producing an interactive HTML report with grades, opportunity themes, and fix prompts. - Guided Editing: Surgically edits existing agents while preserving identity coherence, with cascade assessment across SKILL.md, sanctum templates, and capability references. - Use Case: Ask the builder to create a personal coding coach agent; it determines the agent needs memory, gathers identity and CREED seeds, generates the sanctum templates and init script, and outputs a complete agent skill ready to install. ## Quick Start Ask the agent builder to create a new agent, for example: build an agent that acts as a creative writing coach with memory between sessions.
