remembering-conversations

Searching past conversations lets you retrieve decisions, code patterns, and gotchas from prior work without reloading full chat histories. Dispatch the search-conversations subagent with a semantic query before starting features, debugging, or architectural planning to synthesize relevant findings in 200-1000 words, reducing context usage by up to 100x compared to raw conversation recall.

Apply conversation recall proactively before any non-trivial task: feature implementation, debugging, architectural decisions, or responding to partner references. Semantic search of episodic memory prevents reinventing solutions and surfaces past mistakes, design patterns, and implementation gotchas specific to your project history.

Yes, conversation search supports date filtering to pinpoint historical context within specific timeframes. Combined with semantic and text search modes, date filtering lets you isolate relevant discussions from particular project phases or decision periods.

Semantic search finds conceptually related discussions even when terminology differs, while text search matches exact phrases. The skill supports vector, text, or combined search modes so you can choose precision or concept-matching based on whether you're hunting specific code examples or exploring related design decisions.

Conversation search synthesizes findings from top matches into actionable briefs with sources, consuming 50-100x less context than loading raw histories. Manual review requires scanning full conversations; semantic search delivers decision-support summaries aligned to your current task.

Yes, conversation recall depends on episodic memory containing past conversations and project work. The skill queries your conversation history; without prior recorded sessions, there is no historical data to search. Start using it immediately for all future work to build searchable context over time.