agent-selector

Agent selector routes prompts to specialized agents based on keywords and task type. Provide a prompt like 'write unit tests for auth module' and it automatically selects the tester agent; 'refactor payment logic' routes to the refactor agent, speeding task execution.

Agent selector routes seven task categories: testing, research, code review, refactoring, documentation, cleanup, and feature implementation. Each maps to a specialized agent optimized for that specific work type.

Agent selector uses priority-ordered keyword matching to parse user prompts and identify intent. Keywords like 'test,' 'review,' 'document,' and 'refactor' trigger routing to the corresponding agent, reducing misrouting and ensuring faster, more accurate task completion.

Agent selector analyzes prompt context and keywords to identify the primary task type, then routes to the best-fit agent. For mixed requests, the priority-ordered matching mechanism selects the dominant task type to route appropriately.

Agent selector falls back to context analysis and task type inference. If no clear keyword match exists, it evaluates surrounding prompt context to determine the closest agent fit, ensuring routing even for ambiguous or novel requests.