Crash Recovery And Resilience With Process Supervision

A Claude Code plugin that automatically captures everything Claude does during your coding sessions, compresses it with AI (using Claude's agent-sdk), and injects relevant context back into future sessions.

Unique: Implements multi-layer crash recovery: in-memory queues with periodic SQLite checkpoints, Bun-managed process supervision for automatic restarts, and graceful shutdown handlers that flush queues before termination. On restart, the system replays queued observations from SQLite, ensuring no data loss. This is distinct from systems that rely solely on cloud persistence

vs others: More resilient than in-memory-only systems because observations are persisted to SQLite even if the process crashes; more automatic than manual recovery because Bun restarts the worker service without user intervention; more complete than simple logging because it preserves both queued and processed observations

via “process lifecycle management with automatic respawning”

** - A local MCP server for developers that mirrors your in-development MCP server, allowing seamless restarts and tool updates so you can build, test, and iterate on your MCP server within the same AI session without interruption.

Unique: Couples process lifecycle with proxy session persistence — respawned processes automatically reconnect through the same proxy socket, preserving client context. Uses ProcessManager abstraction to decouple lifecycle logic from proxy forwarding logic.

vs others: More integrated than generic process managers (PM2, systemd) because it understands MCP protocol semantics and coordinates with proxy state; more lightweight than full orchestration platforms.

via “exception-handling-and-recovery”