Described is a technology by which support information (e.g., help and/or other information) related to a computer program is automatically and dynamically searched with respect to a task of that program, with the results of the search aggregated for output in relation to the program. For example, the search may be based on active context data of the task (e.g., such as error data and the current operating environment), so that the support information that is located matches what is likely relevant. User preference data and default data may be used to narrow the search to specific sites, such as community forums that have been deemed to be valuable. Other input used to generate the query or queries may include historical and/or statistical data, and/or explicit user input.