context-synthesis

Efficient context gathering is done by orchestrating parallel retrieval from memory, project specifications, and web searches. This approach filters and structures findings to minimize token usage while maximizing relevance for analysis.

Parallel information synthesis works by simultaneously querying sources like claude-mem, serena memories, existing specs, and web searches. It then consolidates these findings into a structured summary highlighting key insights and their relevance.

The best way to prepare is to gather background information on the project, past decisions from memory, and relevant industry best practices from the web, presenting a concise summary to inform your questions.

Yes, context gathering prioritizes existing project specifications alongside memory sources and web searches. It retrieves information from these sources in a parallel and selective manner to build comprehensive context.

To minimize token usage during context gathering, the process filters and structures findings selectively. It prioritizes sources like claude-mem, serena memories, existing specs, and web searches to maximize relevance without excess tokens.

Parallel context synthesis may not be suitable when you need deep, sequential analysis of a single source rather than consolidated findings from multiple sources, or when token efficiency is not a constraint for your task.