agent-discovery

Agent refactoring involves analyzing agent definitions and prompts to identify responsibilities, tool usage, and opportunities for modular skill extraction. This structural decomposition breaks down monolithic architectures into smaller, focused, and reusable components.

You should extract capabilities into a modular skill library when your complex agent becomes difficult to maintain. Identifying redundant procedures and logic within monolithic prompts indicates it is time to refactor them into reusable skills for better architecture management.

For structural analysis of an agent architecture, you need to provide a readable agent definition or prompt. The analysis parses this input to identify responsibilities, tool usage, and failure points to perform pattern identification and structural decomposition.

The best way to identify reusable procedures in a bloated agent prompt is through structural analysis that inventories responsibilities and tool usage. This process pinpoints multi-step logic that is better suited as standalone skills rather than prompt instructions.

Yes, agent refactoring works without external dependencies because the process relies entirely on analyzing the provided agent definition or prompt. It performs structural decomposition and pattern identification directly on the readable input text.