OpenClaw Docs Audit

Continuous compliance checker for the openclaw-client-bootstrap skill against upstream OpenClaw releases and documentation.

On Trigger

Run the audit script first, then do a deep-dive analysis.

Step 1: Run the audit script

AUDIT="$(ls ~/.claude/skills/openclaw-docs-audit/scripts/audit.sh \
            ~/.codex/skills/openclaw-docs-audit/scripts/audit.sh \
            ./scripts/audit.sh 2>/dev/null | head -1)"
bash "$AUDIT" --instances

This produces a quick structural report: version drift, removed keys, instance staleness, and SKILL.md coverage gaps.

Step 2: Fetch upstream release notes

For each release newer than the pinned version, fetch the full release body:

gh release view <tag> --repo openclaw/openclaw --json body -q '.body'

Scan each release for:

• Config schema changes (new keys, removed keys, type changes, renamed keys)
• Default value changes
• New validation rules (openclaw doctor changes)
• Security fixes that affect config (SSRF, auth, sandbox, exec)
• Channel behavior changes (dmPolicy, groupPolicy, allowFrom enforcement)

Step 3: Fetch upstream config reference

# WebFetch the config reference for the latest schema

Use WebFetch on https://docs.openclaw.ai/gateway/configuration-reference to get the authoritative config schema. Compare against the snapshot in references/config-schema-snapshot.md.

Step 4: Diff and classify changes

Cross-reference upstream changes against THREE targets:

1. Template config (assets/client-kit/openclaw.json)
2. Deployed instance configs (assets/instances/*/openclaw.json)
3. SKILL.md schema notes (the "Config Schema Notes" section)

Classify each finding:

Breaking (must fix before next deploy)

• Removed config keys still present in template/instances
• Type changes that would crash gateway startup
• Security fixes that change default behavior (e.g., dmPolicy enforcement)
• Validation rules that now reject previously-valid configs

Recommended (should fix in next maintenance window)

• New config keys that improve security posture
• Default value changes that alter behavior
• Deprecated keys that still work but will be removed
• Schema notes in SKILL.md that don't cover recent versions

Nice-to-have (opportunistic improvements)

• New features available via config (new channels, tools, plugins)
• Performance tuning knobs added upstream
• UX improvements (streaming modes, button styles, etc.)
• CLI commands that could simplify operations

Step 5: Produce the report

Output a structured report with:

## Breaking Changes (N items)
- [version] Description of what changed and what breaks
  - Affected: template / instance-name / SKILL.md
  - Fix: Concrete config change or SKILL.md update

## Recommended Changes (N items)
- [version] Description
  - Affected: ...
  - Fix: ...

## Nice-to-Have (N items)
- [version] Description
  - Benefit: ...

## Suggested SKILL.md Updates
- Add schema notes for versions X through Y
- Update meta.lastTouchedVersion to Z
- Add/remove keys from removed-keys list
- Update validate_client_kit.sh checks

## Suggested Config Changes
- Template: { specific JSON changes }
- Instance X: { specific JSON changes }

Step 6: Update the schema snapshot

After analysis, update references/config-schema-snapshot.md in THIS skill with any newly discovered schema changes so future audits have a local baseline.

Reference Files

• scripts/audit.sh — Quick structural drift check (version, keys, instances)
• references/config-schema-snapshot.md — Point-in-time schema reference with per-version change notes

Upstream Sources

• Repo: openclaw/openclaw on GitHub
• Releases: gh release list --repo openclaw/openclaw
• Config docs: https://docs.openclaw.ai/gateway/configuration-reference
• Install: curl -fsSL https://openclaw.ai/install.sh | bash

Integration with openclaw-client-bootstrap

After completing the audit, the recommended workflow is:

1. Apply breaking fixes to template + instances
2. Update SKILL.md schema notes section
3. Update meta.lastTouchedVersion in template openclaw.json
4. Run validation: bash scripts/validate_client_kit.sh assets/client-kit
5. Run review: bash scripts/review_kit.sh --skill
6. For live claws: bash scripts/review_kit.sh --live
7. Commit changes to the bootstrap skill repo