OpenClaw Signals: Agent State Recovery, Lifecycle Refactors, and DBus Latency Cuts

PR

PR #330 agent lifecycle refactor for resource handling

This PR overhauls how the agent handles resource lifecycle and cleanup. If you manage deployments at scale, getting these mechanics right means fewer zombie processes and tighter memory footprints.

PR #328: Implementing robust state recovery logic

This PR implements the code to recover lost state on restart, directly addressing issues #1560. It is high-signal work because without this, resilient agent orchestration is just a hope, not a contract.

PR #327: Removing DBus dependency for lower latency

Latency was cut by stripping out some DBus dependency paths in this update. It is a subtle architectural win that removes a layer of IPC overhead and makes the system feel snappier.

Bug

Issue #1559: Default access path configuration is too permissive

A bug report on default path handling was filed. Tightening up file access defaults is a necessary step for hardening, but builders need to audit their integration scripts to ensure nothing breaks.

Issue #1560: State recovery failure during agent restart

Handling restarts has been a pain point recently; this issue tracks the specific failure mode where state is lost silently. If you are running headless agents, watch this fix closely to avoid ghost data.