A protective workflow tool that enforces a plan-then-apply sequence for modifying OpenClaw system configurations.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install oc-guard-skill
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install oc-guard-skill using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is oc_guard?
oc_guard is a critical management skill designed to act as a security layer for your agent environment. Instead of allowing direct, error-prone edits to core configuration files like openclaw.json, oc_guard mandates a structured process of planning and explicit confirmation. As a foundational piece of the Openclaw Skills ecosystem, it ensures that every change is validated and documented through bilingual execution receipts, maintaining the stability of your AI coding environment.
This skill is particularly valuable for developers who need to maintain complex agent setups without risking manual corruption of settings. By utilizing the bundled Python scripts, oc_guard provides a robust interface for both human operators and automated processes to interact with system settings safely. It is an essential utility for anyone looking to scale their use of Openclaw Skills while adhering to best practices in configuration management.
oc_guard Use Cases
- Updating API keys and model provider endpoints without direct file manipulation.
- Staging complex configuration changes using proposal files for review.
- Preventing accidental system-wide configuration breaks during automated updates.
- Generating audit-friendly bilingual receipts for every configuration change made to Openclaw Skills.
How oc_guard Works
- The user or agent initiates a request to change a setting via the plan command.
- oc_guard analyzes the requirement and generates a detailed proposal of the changes.
- The system returns a bilingual execution receipt to the user for validation.
- If the change is high-risk, the user must explicitly run the apply command with a confirmation flag.
- The skill updates the underlying openclaw.json file and provides a final success receipt.
oc_guard Setup
To get started with this component of your Openclaw Skills library, ensure you have Python 3 and the OpenClaw core installed. Follow these steps:
# Navigate to the skill directory
cd {baseDir}/scripts/
# Ensure the guard script is executable
chmod +x oc-guard.py
# View available commands and options
./oc-guard.py --help
oc_guard Data Schema & Taxonomy
| Data Component | Description |
|---|---|
| Config Path | Primary target is ~/.openclaw/openclaw.json |
| Proposal Files | Markdown or JSON files used to stage bulk configuration changes |
| Receipt Markers | Specific bilingual tags like 【模型说明-未执行】 for tracking execution status |
| Environment | Requires python3, openclaw, and opencode binaries |
oc_guard Advanced Features
- Support for --proposal flags to handle complex, multi-layered configuration updates from external files.
- Mandatory --confirm flags for high-risk operations to ensure human-in-the-loop safety.
- Bilingual execution receipt system designed for global development teams using Openclaw Skills.
- Deep integration with the OpenClaw CLI for seamless workflow transitions between planning and application.
SKILL.md
Loading
Related Openclaw Skills
Kokoro TTS
A high-performance local text-to-speech skill for generating realistic AI voices via simple CLI commands.
YouTube Transcriber
A high-performance command-line tool that automates YouTube video transcription using existing subtitles or OpenAI Whisper AI.
Reddit Content Assistant
An AI-driven strategist that helps developers craft authentic Reddit content and track community engagement through data-backed workflows.
Protea: Self-Evolving Life Agent
Protea is a three-ring autonomous agent architecture that leverages LLMs to self-restructure, self-reproduce, and self-evolve its own source code.
AOI Council
A template-driven decision synthesis tool that uses multiple expert perspectives to stress-test plans and ideas.
AOI Cron Ops (Lite)
A specialized diagnostic tool for auditing, optimizing, and controlling costs for OpenClaw cron jobs through detailed reporting.
METADATA
Featured*
