ops-framework

Coordinate long-running read tasks with checkpointing and stall detection.

Updated Mar 1, 2026
One-click install
npx skills add https://github.com/vuxsoul-ui/catclash --skill ops-framework
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ops-framework
Source: https://github.com/vuxsoul-ui/catclash/tree/main/skills/ops-framework
Command: npx skills add https://github.com/vuxsoul-ui/catclash --skill ops-framework

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

OpenClaw's Ops Framework provides a zero-token capability to run and monitor long-running read tasks with checkpoint/resume, safe stall detection, and alerting, while ensuring write operations are blocked by default until explicit approval.

Core Features & Use Cases

  • Long-running read tasks: checkpointing, pause/resume, stall detection, and periodic progress reporting.
  • One-shot reads: health checks and quick verifications that report only on ACTION REQUIRED or ALERT.
  • One-shot writes: blocked by default; require explicit approval and verification chains to a read-only task.
  • Telegram alerting: realtime Ops快报 for exceptions and progress.

Quick Start

Copy the OPS_FRAMEWORK files to your OpenClaw host and run the local ops-monitor.py with your config.

Frequently Asked Questions about ops-framework

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I monitor long-running read tasks with checkpoint and stall detection?

You can monitor long-running read tasks using checkpointing, pause/resume, and stall detection to safely track progress and resume operations after interruptions without losing data state.

How do I send Telegram alerts for task exceptions and progress reporting?

Telegram alerting sends realtime Ops快报 notifications for exceptions and periodic progress reporting during long-running task execution, ensuring you stay informed of critical failures without manual polling.

Can I execute one-shot write operations without explicit approval?

One-shot writes are blocked by default. They require explicit approval and a verification chain linked to a read-only task to ensure sensitive operations are safely controlled.

What is the best way to run health checks and quick verifications without token overhead?

Use the zero-token one-shot read capability for health checks and quick verifications, which reports exclusively on ACTION REQUIRED or ALERT statuses to minimize overhead.

Are there quiet hours or safety policies for task scheduling?

Yes, the framework enforces safety policies including configurable quiet hours, read-only reads by default, write-blocking, and explicit approval workflows to protect sensitive operations during synchronization.