context-recovery

To restore lost project context after an LLM memory compaction, you can trigger a recovery pass to rehydrate the workspace. This reads a CONTEXT-RECOVERY.md snapshot and daily memory logs to reconstruct the active project, latest task, and next steps.

The best way to resume an agent session after a context reset is to use an automated context recovery mechanism. It reads recent memory diary entries and uses a memory_search utility to surface relevant project state before presenting a concise summary.

You need to update the CONTEXT-RECOVERY.md snapshot at session milestones during long-running conversations or multi-step workflows. This keeps the project state, last task, and next steps fresh for recovery on the next run.

Yes, context recovery works for multi-project workspace environments. The structured restoration process applies across single-project workspaces, multi-project contexts, and when agent sessions resume after sleep.

To rehydrate a workspace from daily memory logs, instruct the agent to perform a recovery pass. The system reads the logs alongside a structured CONTEXT-RECOVERY.md snapshot to rebuild the active project state and repository status.

Your LLM loses critical project state during long-running conversations because memory compaction discards older context. A deterministic recovery mechanism solves this by reconstructing the active project, last task, next steps, and repository status from saved snapshots.