What problem does it solve? Defining a reusable custom agent from scratch is hard: users must decide the persona, tool restrictions, and scope, then encode them correctly in an .agent.md file. This Skill guides that process by extracting specialization patterns from the current conversation and turning them into a working agent definition. ## Core Features & Use Cases - Conversation-Based Extraction: Reviews the conversation history to generalize an emergent specialized role, tool preferences, and domain scope into a custom agent. - Guided Clarification: Asks targeted questions about the agent's job, when it should be chosen over the default agent, and which tools it should use or avoid. - Iterative Refinement: Drafts the .agent.md file, identifies ambiguous or weak sections, and iterates with the user until finalized. - Use Case: After spending a session constraining the assistant to only review Python security issues, use this Skill to capture that behavior as a reusable security-reviewer agent. ## Quick Start Ask the assistant to create a custom agent for the specialized workflow you have been using in this conversation.