config-guardian

What This Skill Does

Config Guardian acts as the defensive layer for your OpenClaw agent, ensuring that any modifications to the openclaw.json configuration file are performed safely and predictably. It eliminates the risk of introducing syntax errors, invalid configuration keys, or unsupported values that could otherwise cause the OpenClaw gateway to crash or behave unexpectedly. By leveraging the atomic_apply.sh workflow, the skill enforces a strict process: it creates a timestamped backup in ~/.openclaw/config-guardian-backups/, applies the requested change, and immediately runs a diagnostic check using 'openclaw doctor'. If the validation fails, the skill automatically triggers a rollback to the previous known-good state, protecting your agent's stability.

Installation

To integrate Config Guardian into your agent's toolset, execute the following command in your terminal: clawhub install openclaw/skills/skills/abdhilabs/config-guardian Once installed, the agent will have access to the atomic_apply script and associated utilities necessary for managing system configurations.

Use Cases

This skill is indispensable for automated environment management. Use it when:

• Updating the primary AI model for agent tasks via configuration settings.
• Modifying API endpoint configurations or gateway routing rules.
• Updating system-wide performance tuning variables.
• Managing agent personality or behavior parameters stored within the main JSON configuration.
• Ensuring zero-downtime updates in production-critical agent environments.

Example Prompts

1. "OpenClaw, update the primary model to minimax-portal/MiniMax-M2.5 using Config Guardian."
2. "Please change the log level in the config file to 'verbose' and ensure the change is validated."
3. "Adjust the gateway timeout setting to 3000ms using the atomic apply workflow to prevent breaking the system."

Tips & Limitations

• Always prioritize the atomic_apply.sh script for all modifications; manual editing of the JSON file is strongly discouraged when this skill is available.
• While the tool provides robust validation, it cannot prevent logical errors if you supply a valid-type value that is semantically incorrect for your current workflow.
• Remember that the agent must have write permissions for the configuration directory.
• If a validation error occurs, do not attempt to force the update; investigate the provided logs to understand why the doctor check failed before re-attempting the change.