Memory Continuity: Zero-Dependency Session Persistence for AI Agents

The Problem Every time your AI coding agent hits a session reset — timeout, context limit, gateway restart — it forgets everything. Your project context, decisions made, files touched, open questions... gone. The Solution Memory Continuity is a zero-dependency lifecycle plugin that preserves working state across session boundaries using plain markdown files. No SQLite. No vector DB. No external APIs. It answers one question: What were we doing, where did we stop, and what should happen next? How It Works 6 lifecycle hooks capture state at critical moments: before_agent_start — inject relevant history into new sessions before_compaction — save state before context compression before_reset — checkpoint before /clear agent_end / session_end — persist everything at shutdown subagent_ended — recover child agent results into parent All data lives in a memory/ directory as plain .md files: memory/ CURRENT_STATE.md # live working state sessions/2026-03-31.md # daily session logs summaries/dail