How do I install the Openclaw Config Guard skill?

Run npx skillmds@latest add zerone-agent/openclaw-config-guard in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Openclaw Config Guard safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Openclaw Config Guard?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Openclaw Config Guard free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Openclaw Config Guard?

zerone-agent (@zerone-agent) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Openclaw Config Guard by zerone-agent

Name: Openclaw Config Guard
Author: zerone-agent

Openclaw Config Guard

Audit and safely repair OpenClaw configuration with deterministic validation, backups, rollback, and change reporting. Use when asked to review or modify `openclaw.json`, check whether OpenClaw can still start, safely fix startup-blocking config errors, or audit OpenClaw config before deciding on changes.

zerone-agent cc828a2 3 files · 19.2 KB Updated Sep 17, 2026

File contents

OpenClaw Config Guard

Audit first. Repair only when the fix is proven. Protect startup over aesthetics.

Required Sources

Before making any judgment, open the official docs listed in references/official-sources.md. Treat them as the source of truth for schema, allowed values, and repair guidance. Do not rely on memory for config rules.

Workflow

Resolve the active config path:

python3 "<skill-dir>/scripts/config_guard.py" resolve-path --json

If that fails, fall back to ~/.openclaw/openclaw.json.

Run a deterministic audit before touching the file:

python3 "<skill-dir>/scripts/config_guard.py" audit --doctor

This wraps:

openclaw config validate --json
optional openclaw doctor --non-interactive

Classify findings:

startup blockers: JSON5 parse failures, schema validation failures, unknown keys, wrong types, invalid enum values, missing required structure, or clearly conflicting settings that prevent startup.
recommendations: suspicious but non-blocking items such as duplicate plugin IDs, stale-but-working config, style cleanup, or non-critical hardening suggestions.

Decide whether you may auto-fix:

Only auto-fix if the issue is a startup blocker.
Only auto-fix if the docs or CLI output clearly show the correct repair.
Prefer openclaw config set / openclaw config unset for exact path edits.
Use manual JSON5 edits only when the CLI cannot express the required change and preserving comments or structure matters.
Never run openclaw doctor --fix by default.
Never restart OpenClaw by default.

Backup before any write:

python3 "<skill-dir>/scripts/config_guard.py" backup --json

Re-validate after any write:

python3 "<skill-dir>/scripts/config_guard.py" validate --doctor --json

If post-change validation fails, roll back immediately from the backup and say so in the report.

Summarize what changed:

python3 "<skill-dir>/scripts/config_guard.py" diff --before /path/to/before --after /path/to/after --json

If you want a deterministic report frame, prepare a JSON manifest and run:

python3 "<skill-dir>/scripts/config_guard.py" report --manifest /path/to/manifest.json

<skill-dir> means the directory that contains this SKILL.md. Resolve relative paths against this skill directory instead of assuming any environment variable is set.

Decision Boundaries

Do not change non-blocking issues without user approval.
Do not guess undocumented keys or values.
Do not rewrite the whole config just to normalize formatting.
Do not claim success without rerunning validation.
Do not leave the user without a backup path, modified paths list, and post-change validation result.

Report Requirements

The final Markdown report must include:

official sources consulted
active config path
pre-change validation result
startup blockers found
automatic fixes applied
issues intentionally not auto-fixed and why
non-blocking recommendations for user decision
modified config paths
backup path
post-change validation result
whether manual restart is needed, and why

Resources

Official source list: references/official-sources.md
Deterministic helper script: scripts/config_guard.py

---name: openclaw-config-guarddescription: Audit and safely repair OpenClaw configuration with deterministic validation, backups, rollback, and change reporting. Use when asked to review or modify `openclaw.json`, check whether OpenClaw can still start, safely fix startup-blocking config errors, or audit OpenClaw config before deciding on changes.---# OpenClaw Config GuardAudit first. Repair only when the fix is proven. Protect startup over aesthetics.## Required SourcesBefore making any judgment, open the official docs listed in [references/official-sources.md](references/official-sources.md). Treat them as the source of truth for schema, allowed values, and repair guidance. Do not rely on memory for config rules.## Workflow1. Resolve the active config path:```bashpython3 "<skill-dir>/scripts/config_guard.py" resolve-path --json```If that fails, fall back to `~/.openclaw/openclaw.json`.2. Run a deterministic audit before touching the file:```bashpython3 "<skill-dir>/scripts/config_guard.py" audit --doctor```This wraps:- `openclaw config validate --json`- optional `openclaw doctor --non-interactive`3. Classify findings:- `startup blockers`: JSON5 parse failures, schema validation failures, unknown keys, wrong types, invalid enum values, missing required structure, or clearly conflicting settings that prevent startup.- `recommendations`: suspicious but non-blocking items such as duplicate plugin IDs, stale-but-working config, style cleanup, or non-critical hardening suggestions.4. Decide whether you may auto-fix:- Only auto-fix if the issue is a startup blocker.- Only auto-fix if the docs or CLI output clearly show the correct repair.- Prefer `openclaw config set` / `openclaw config unset` for exact path edits.- Use manual JSON5 edits only when the CLI cannot express the required change and preserving comments or structure matters.- Never run `openclaw doctor --fix` by default.- Never restart OpenClaw by default.5. Backup before any write:```bashpython3 "<skill-dir>/scripts/config_guard.py" backup --json```6. Re-validate after any write:```bashpython3 "<skill-dir>/scripts/config_guard.py" validate --doctor --json```If post-change validation fails, roll back immediately from the backup and say so in the report.7. Summarize what changed:```bashpython3 "<skill-dir>/scripts/config_guard.py" diff --before /path/to/before --after /path/to/after --json```If you want a deterministic report frame, prepare a JSON manifest and run:```bashpython3 "<skill-dir>/scripts/config_guard.py" report --manifest /path/to/manifest.json````<skill-dir>` means the directory that contains this `SKILL.md`. Resolve relative paths against this skill directory instead of assuming any environment variable is set.## Decision Boundaries- Do not change non-blocking issues without user approval.- Do not guess undocumented keys or values.- Do not rewrite the whole config just to normalize formatting.- Do not claim success without rerunning validation.- Do not leave the user without a backup path, modified paths list, and post-change validation result.## Report RequirementsThe final Markdown report must include:- official sources consulted- active config path- pre-change validation result- startup blockers found- automatic fixes applied- issues intentionally not auto-fixed and why- non-blocking recommendations for user decision- modified config paths- backup path- post-change validation result- whether manual restart is needed, and why## Resources- Official source list: [references/official-sources.md](references/official-sources.md)- Deterministic helper script: [scripts/config_guard.py](scripts/config_guard.py)

Frequently asked questions