Periodically ask assistant to recall everything it knows about the user

User repeatedly (in different sessions, often at session start or in plan mode) asks the assistant identity/memory-recall questions like 'Who am I?' or 'Tell me everything you know about me.' This is used to check long-term memory accuracy and surface context (identity, relationships, employer, projects, work preferences). When this happens: synthesize a comprehensive, organized summary from long-term memory covering identity/emails, relationships and interpersonal context (especially Sentry colleagues like David Cramer, Dave Rosenthal, Michi), current projects (Lore, opencode config, Sentry work, home infra), and known working-style preferences (objectivity over validation, root-cause-first, persistent fixes, verification against source of truth, harsh code review discipline, prose style rules). Explicitly flag that the memory may be stale and offer to verify specific threads against current source/system state rather than assuming memory is fully accurate.