What problem does it solve? AI agent sessions lose context between runs, forcing users to re-explain prior work. This Skill provides a persistent file-based backend so agents can save their state at session end and reload it at the next session start. ## Core Features & Use Cases - Save Operation: Writes a compiled agent state block (Wrapped, Context, Plan, Next, Learned, Spec Sync fields) to progress/[agent_id].md at the project root. - Load Operation: Reads back the stored state file for a given agent, returning nothing when no prior state exists so callers can handle fresh starts. - Use Case: A developer wraps up a work session with the saga agent; the wrap tool calls this Skill to persist the session state, and the next day's start tool reloads it to resume exactly where work stopped. ## Quick Start Ask the agent to save its current session state to the project progress folder so it can be restored at the next session start.