What problem does it solve? AI agents lose context between sessions, causing repeated mistakes and forgotten decisions. This Skill defines a strict discipline for when and how to persist decisions, bugfixes, and discoveries into Engram-style persistent memory so knowledge survives across sessions and compactions. ## Core Features & Use Cases - Save Rules: Triggers mem_save immediately after decisions, bugfixes, pattern discoveries, and preference changes, using structured What/Why/Where/Learned content and stable topic keys. - Search Rules: Mandates mem_context before mem_search on recall requests, and proactive mem_search before similar work or when a user references a project or problem. - Session Close Rules: Requires mem_session_summary with goal, discoveries, accomplishments, next steps, and relevant files before ending a session, plus a recovery procedure after compaction. - Use Case: After fixing a non-obvious production bug, the agent immediately saves the root cause and fix location to memory, then recalls it automatically weeks later when a similar bug report arrives. ## Quick Start Apply the Engram memory protocol so every decision, bugfix, and discovery in this session is saved and searchable in future sessions.
