boot-resume
Zero-cooperation session recovery after gateway restart. No checkpoints, no hooks, no agent involvement — just reads the evidence and picks up where it left off. Use when: the gateway was killed mid-task (SIGTERM, OOM, SIGKILL, crash), sessions were interrupted mid-turn with tool calls in progress, the agent stopped responding after a restart, a user reports the agent went silent after a crash, you need to manually check whether any sessions need recovery, or you want automatic resume without writing any checkpoint logic.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/belugary/boot-resumeWhat This Skill Does
The boot-resume skill is a critical recovery mechanism for OpenClaw gateway instances. Designed as a zero-cooperation recovery tool, it allows the agent to recover from abrupt shutdowns without requiring explicit checkpoint logic or agent-side intervention. When a gateway undergoes a hard restart, crash, or OOM event, it often leaves active, mid-task agent sessions in a 'zombie' state. Boot-resume solves this by scanning session history files, identifying the exact point of interruption based on the last JSONL entries, and automatically triggering a wake event to resume the task. It operates entirely outside the agent's logic, using a system-level shell script that inspects log state to determine if an assistant was mid-execution or awaiting a tool result.
Installation
Installation is streamlined for immediate deployment. You can use the automated command: bash {baseDir}/install.sh. This will deploy the necessary check script to your workspace, configure the systemd drop-in file to run on gateway boot, and set up the systemd user service for wake/suspend events. For manual control, copy the boot-resume-check.sh to your workspace scripts directory, apply the executable permissions, and move the service templates to your user configuration folder before executing systemctl --user daemon-reload and enabling the wake service.
Use Cases
This skill is essential in high-uptime environments where gateway stability is non-negotiable. Use cases include recovering from SIGTERM or SIGKILL signals that occur during heavy task execution, restoring agent state after an OOM (Out-of-Memory) event, or ensuring that long-running automations successfully complete even if the underlying infrastructure is cycled. It is particularly useful for unattended agents that must survive host reboots or power cycles without manual intervention.
Example Prompts
- "OpenClaw, did you recover the interrupted task after the last system reboot?"
- "Check the current session status to see if any pending tool results were caught by boot-resume."
- "Verify that the agent resumed the data scraping task correctly following the gateway crash yesterday."
Tips & Limitations
Boot-resume uses a 20-minute lookback window by default, meaning it only detects issues that occurred recently. If your gateway is down for longer than 20 minutes, you may need to adjust the scan window in the script. The system relies on the integrity of JSONL session files; ensure that your storage medium is mounted correctly at boot for successful detection. Because this tool interacts directly with the OpenClaw system internals, monitor the /tmp/openclaw/boot-resume.log file to verify that the detection rules are triggering correctly in your specific environment.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-belugary-boot-resume": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-read, file-write, code-execution