openclaw-config-guardian
Skill2 starsChanged 8 months agoMITView on GitHub
Standardize audit/patch `~/.openclaw/openclaw.json` process after OpenClaw update (schema/docs/runtime drift), with mandatory backup, verify and rollback.
In this repository47 other files it gives its agents.
- brainstormingSkill
- claude-code-termux-runbookSkill
- cli-patcherSkill
- code-review-workflowSkill
- commit-commandsSkill
- consulting-analysisSkill
- context7-cleanSkill
- data-analysisSkill
- deep-researchSkill
- edit-tool-compat-guardSkill
- env-managerSkill
- executing-plansSkill
- feature-devSkill
- firecrawlSkill
- frontend-designSkill
- gateway-device-pairingSkill
- githubSkill
- glibc-prebuilt-bundleSkill
- gogSkill
- image-generationSkill
- memory-setupSkill
- module-auditorSkill
- module-scaffoldSkill
- openclaw-ollama-runtime-kbSkill
- plugin-cutover-guardianSkill
- plugin-runtime-guardianSkill
- python-env-managerSkill
- self-improving-agentSkill
- simplify-and-hardenSkill
- skill-benchSkill
- skill-creatorSkill
- skill-unificationSkill
- skill-vetterSkill
- socratic-groundingSkill
- subagent-orchestrationSkill
- systematic-debuggingSkill
- tailscale-oauth-bridgeSkill
- termux-packagerSkill
- termux-runnerSkill
- test-driven-developmentSkill
- vercel-deploy-claimableSkill
- verification-before-completionSkill
- video-generationSkill
- workspace-cleanupSkill
- writing-plansSkill
- yt-dlp-ingest-firstSkill
- zero-downtime-cutoverSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
18 rules. About 0 are common, 16 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈16
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 36
Backup first: always create a backup before writing config.
- Distinctive: line 37
Minimum patch: only correct approved items, do not touch logic outside scope.
- Distinctive: line 38
Verify after patch: always include a re-check step with CLI.
- Distinctive: line 39
Rollback-ready: if verification fails, restore the backup immediately.
- Distinctive: line 64
Read local docs to confirm current fields, avoid guessing the schema.
- Distinctive: line 65
If there is a conflict between runtime behavior and docs: prioritize actual runtime + doctor output, and state it clearly in the report.
- Distinctive: line 73
Only apply classes approved by Boss.
- Distinctive: line 79
Do not delete keys unless there is a clear instruction or doctor requests fixing unknown keys.
- Distinctive: line 82
After patching, run at minimum:
- Distinctive: line 89
no new errors arise due to the patch
- Distinctive: line 92
If verification fails because of the patch:
- Very specific: line 111
Do not treat runtime warnings (pairing/OAuth/service supervisor) as schema errors if config parsing is still valid.
- Distinctive: line 112
Do not arbitrarily touch tokens/secrets without approval.
- Very specific: line 113
Do not run doctor --fix automatically in production without approval (because it may drop unknown keys).
- Distinctive: line 114
When lacking schema evidence: stop at the report + request a specific patch scope.
- Distinctive: line 137
If input/permissions are missing: report what is missing and clearly request additional information.
- Distinctive: line 138
If runtime/tool error: brief error return + possible cause + suggested fix.
- Distinctive: line 139
If there is a risk policy/safety concern: stop and escalate according to the approval checkpoint.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
OpenClaw Config Guardian
Indexed sub-skill references (source map)
- @[SKILL.md#L1-L136] Standard Frontmatter (
name,description). - @[SKILL.md#L6-L136] OpenClaw Config Guardian.
- @[SKILL.md#L8-L136] Target.
- @[SKILL.md#L11-L136] When to use.
- @[SKILL.md#L17-L136] Required rule.
- @[SKILL.md#L24-L136] Standard path.
- @[SKILL.md#L33-L136] Standard Procedure (SOP).
- @[SKILL.md#L80-L136] Standard report template.
- @[SKILL.md#L93-L136] Error-proof guardrails.
- @[SKILL.md#L99-L136] Quick checklist before closing.
- @[SKILL.md#L106-L136] Expected result.
Scope
- In-scope: audit/patch
~/.openclaw/openclaw.jsonafter an OpenClaw update, backup-first, verify-first, rollback-ready. - Out-of-scope: runtime issues that are not config/schema drift, or secret mutations without approval.
Target
Standardize operations on ~/.openclaw/openclaw.json when OpenClaw has just updated its version (schema/docs/runtime changes), reducing incorrect patch errors or config drift across future sessions.
When to use?
Activate this skill when one of the following signs appears:
- Boss just updated OpenClaw and asked to "check/patch config".
- It is suspected that the schema has changed, old fields have been deprecated, or the docs have changed.
openclaw doctor/openclaw status --allreports a warning related to config/gateway/session policy.
Mandatory principles
- Local docs first: prioritize local docs in the module before concluding.
- Backup first: always create a backup before writing config.
- Minimum patch: only correct approved items, do not touch logic outside scope.
- Verify after patch: always include a re-check step with CLI.
- Rollback-ready: if verification fails, restore the backup immediately.
Standard path
- Config:
/data/adb/openclaw/home/.openclaw/openclaw.json - Docs root:
/data/adb/openclaw/node/lib/node_modules/openclaw/docs - Priority references:
docs/gateway/configuration-reference.mddocs/gateway/configuration.mddocs/cli/doctor.mddocs/gateway/troubleshooting.md
SOP
Step 0 — Snapshot runtime
Run:
openclaw statusopenclaw doctoropenclaw status --all(when in-depth diagnosis is needed)
Collect:
- runtime version
- config-related error/warn
- gateway/auth state (if available)
Step 1 — Compare docs/schema
- Read local docs to confirm current fields, avoid guessing the schema.
- If there is a conflict between runtime behavior and docs: prioritize actual runtime + doctor output, and state it clearly in the report.
Step 2 — Create a patch plan by layer
Sort patches before touching files:
- A (Critical/required): metadata version drift, required fields according to the new schema.
- B (Recommended/safe): explicit policy keys (e.g. session scope), slight hardening.
- C (Out-of-scope/runtime issue): gateway pairing, OAuth dir, service state (not considered schema patch).
Only apply classes approved by Boss.
Step 3 — Backup + patch
- Create timestamped backup:
openclaw.json.bak.patch-YYYYMMDD-HHMM
- Record the patch deterministically (idempotent, no meaningless reorder).
- Do not delete keys unless there is a clear instruction or doctor requests fixing unknown keys.
Step 4 — Required verification
After patching, run at minimum:
openclaw config get <each path just edited>openclaw doctoropenclaw status --all(if there was a gateway/auth warning before)
The result must be:
- the new value is correct
- no new errors arise due to the patch
Step 5 — Rollback if failed
If verification fails because of the patch:
- restore the backup immediately
- rerun
openclaw doctor - report failure + cause + new proposed patch
Standard report template
⚡ [patch] scope=A+B3 → done
- changed: <path1>: <old> -> <new>
- changed: <path2>: <old> -> <new>
- backup: <backup_path>
- verify:
- openclaw config get <path>: <value>
- doctor: <summary>
- status --all: <summary>
🛡️ [WARN] <runtime issues not caused by patch>Error-proof guardrails
- Do not treat runtime warnings (pairing/OAuth/service supervisor) as schema errors if config parsing is still valid.
- Do not arbitrarily touch tokens/secrets without approval.
- Do not run
doctor --fixautomatically in production without approval (because it may drop unknown keys). - When lacking schema evidence: stop at the report + request a specific patch scope.
Quick checklist before closing
- There is a backup file
- There is an explicit diff by path
- Each edited key is verified
- Doctor shows no new errors caused by the patch
- There is a noted runtime issue outside config scope if applicable
Expected results
Any session that activates this skill will patch openclaw.json through a consistent process, with backup, verification, and rollback, reducing the risk of incorrect schema when OpenClaw changes versions or docs.
Trigger Rules
- Positive: Triggers when user request matches
openclaw-config-guardianaccording to description: Standardizes audit/patch~/.openclaw/openclaw.jsonprocess after updating OpenClaw (schema/docs/runtime drift), with mandatory backup, verify and rollback. - Negative: Not enabled for requests outside scope or without the workflow/expertise of this skill.
- Escalation: If there is an external action/network write or sensitive operation, an approval checkpoint must be added before executing.
Output Contract
- Deliverable: target tracking result of
openclaw-config-guardian; specify the artifact/path/steps performed (if any). - Validation: reports
PASS/FAILstatus and the main checks that were run. - Failure mode: when there is an error, the root cause + proposed fix must be stated.
Error Handling
- If input/permissions are missing: report what is missing and clearly request additional information.
- If runtime/tool error: brief error return + possible cause + suggested fix.
- If there is a risk policy/safety concern: stop and escalate according to the approval checkpoint.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

