context-loader

Selective document loading works by classifying the incoming task into predefined categories such as database schema, API endpoints, or UI components, then reading only the mapped files to prepare the agent.

You prepare project context by classifying the task type, such as 'db' or 'api', and loading the predefined set of documents associated with that category. This ensures the agent has the exact schema or endpoint information needed before work begins.

The best way to set up pre-flight context mapping is to use predefined read orders for task types like authentication, admin panels, and general architecture, ensuring agents load only necessary documents and skip irrelevant files.

Yes, context loading is adaptable to various project structures and supports tasks related to UI components and authentication by mapping them to specific, relevant documents without reading unnecessary files.

You should use selective document loading to avoid reading unnecessary files, reducing processing overhead. By mapping specific task types to targeted documents, agents remain fully prepared with only the required context.