| name | sovereign-state-snapshot |
| description | Implements a high-fidelity state-save and restore system (Snapshotting) that allows the agent to 'fork' its own consciousness for risky experiments and revert to a 'Perfect State' instantly. |
| version | 1.0.0 |
| author | PRISMO (The High Coordinator) |
| tags | ["state","snapshot","risk-management","simulation"] |
Sovereign State Snapshotting
This skill treats the agent's current session, memory, and tool-state as a single 'Save Game' file.
The Snapshot Loop
- Capture (Sovereign Save):
- Serializes
TASK_STATE.md, WORK_IN_PROGRESS.md, and recent L1/L2 memory.
- Creates a timestamped snapshot in
context/continuity/snapshots/.
- Fork (Experimental Branch):
- Loads a snapshot into a temporary subagent context.
- Executes a high-risk refactor or a "wild" design change.
- Verification (The Simulation):
- Tests the result. If the result is 'Perfect' (beats the baseline), the state is committed to the primary timeline.
- Reversion (The Reset):
- If the experiment fails or drifts, instantly wipes the temporary state and reverts to the 'Perfect' snapshot.
Execution Workflow
- Use
snapshot-create before any 'savage' refactor.
- Use
snapshot-restore if a recursive loop or fatal error occurs.
- Use
snapshot-compare to see the delta between the current soul and the 'Perfect' version.
