ops-framework

What This Skill Does

The ops-framework is a specialized, 0-token automation engine designed to manage long-running tasks and operational monitoring within OpenClaw. By shifting heavy-duty monitoring and task execution from the model's active token stream to a local background script, users can maintain state, track progress, and receive real-time alerts without incurring high compute costs. The framework consists of a robust CLI monitor (ops-monitor.py) and a declarative job configuration (ops-jobs.json), enabling seamless checkpointing, stalling detection, and safe execution workflows.

Installation

1. Ensure your OpenClaw environment is set up. Navigate to your configuration directory.
2. Execute the installation via the OpenClaw hub: clawhub install openclaw/skills/skills/zjianru/ops-framework.
3. Map your OpenClaw state directory to OPENCLAW_HOME.
4. Copy your configuration files to ~/.openclaw/net/config/ops-jobs.json.
5. Validate your configuration by running python3 ops-monitor.py validate-config --config-file ~/.openclaw/net/config/ops-jobs.json to ensure syntax integrity.

Use Cases

• Continuous Monitoring: Keep tabs on long-running data syncs or inventory scans that run for hours. The system monitors process PID, progress keys, and status, alerting you via Telegram only when anomalies or 'ACTION REQUIRED' tags are triggered.
• Automated Health Checks: Schedule periodic linting or service health checks where the system remains silent during normal operation and only notifies you upon failure or warning-level output.
• Safe Write Execution: Execute sensitive system tasks by defining them as one_shot_write jobs. These remain locked by default and require an explicit human-in-the-loop approval before execution, coupled with a mandatory verification step to ensure the integrity of the write operation.

Example Prompts

• "Ops-framework, list the status of all current long-running read jobs and check if any have stalled."
• "Start the sync_inventory job and notify me on Telegram only if the job encounters an error or reaches a critical state."
• "Force-stop the current database_backup job and provide a summary of the progress before termination."

Tips & Limitations

• Efficiency: Leverage the 0-token design by setting up crons/systemd timers to call ops-monitor.py tick. This keeps your OpenClaw context window free for actual intelligence tasks.
• Contract Adherence: Ensure your status commands output strictly formatted JSON. If the output schema breaks, the monitor cannot track progress or detect stalls.
• Safety: Never modify the one_shot_write protection logic. Treat these jobs as read-only templates until a manual security review is performed.