openclaw-ops
Safe OpenClaw gateway configuration operations. Use when reading, analyzing, or modifying openclaw.json. Enforces schema validation + official documentation verification before every change.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/astralwaveorg/openclaw-config-opsOpenClaw Ops
Safe, documented, schema-validated gateway configuration operations.
When to Use
- Any interaction with
openclaw.json: read, analyze, modify config.get,config.schema.lookup,config.patch,config.apply- Adding/removing models, plugins, channels, agents, MCP servers, tools config
- Troubleshooting config validation errors
- Analyzing current config for optimization opportunities
Two Modes
Determine mode before starting any config work.
Mode A: Analysis Only (no config change)
Trigger: user asks to check, review, compare, optimize, or analyze config, without providing a specific target value.
Flow:
config.schema.lookupfor relevant keys- Official doc verification (see Step 2)
- Present findings with schema + doc evidence
- Stop. Wait for user instruction.
Output format:
## Config Analysis
**Finding**: what was found
**Current state**: relevant config values (from config.get)
**Schema evidence**: what schema.lookup confirmed
**Doc evidence**: source URL or QMD reference
**Recommendation**: suggested change (if any), or "no action needed"
Transitioning to Mode B: if user responds with explicit approval to execute a recommended change (e.g. "do it", "apply", "yes, modify it"), switch to Mode B. Start from Step 3 (Form Change Proposal) — Steps 1 and 2 evidence from Mode A can be reused.
Mode B: Modification (config change)
Trigger: user explicitly requests a change AND provides sufficient detail (target key + intended value or clear intent). Examples:
- "add model X with id Y"
- "set thinkingDefault to medium"
- "remove the minimax plugin"
- Responding to a Mode A recommendation with "do it" / "apply" / "确认修改"
Insufficient for Mode B (fall back to Mode A + ask):
- "帮我改一下配置" without specifying what to change
- "优化一下" without target values
- Vague instructions without concrete config keys or values
Flow: all 6 steps below, in strict order. No step may be skipped.
Mode Switching Rules
| Situation | Action |
|---|---|
| Mode A analysis done, user says "do it" | → Mode B, start from Step 3 (reuse Steps 1-2) |
| Mode B Step 3 proposal rejected | → Back to Mode A, re-analyze |
| User intent unclear | → Mode A, ask for clarification |
| Emergency config fix needed | → Still Mode B, but note "emergency" in log |
Mandatory Process — Mode B (6 Steps)
Step 1: Schema Lookup
config.schema.lookup(path="target.config.key")
Verify:
- Key exists in schema (not invented)
- Type matches intended value
enumvalues present → only use values from the enumadditionalProperties: false→ no extra sub-fields allowed- Free-form objects (
additionalProperties: {}) → still verify field names from docs
Schema has key + config doesn't: this means the key can be created. Proceed to Step 2. Schema lookup fails or key doesn't exist: STOP. Do not proceed. Report to user.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-astralwaveorg-openclaw-config-ops": {
"enabled": true,
"auto_update": true
}
}
}Related Skills
openclaw-ref
OpenClaw 配置、CLI、排错、模型管理完整参考手册。修改 openclaw.json、执行 CLI 命令、排查问题时必须查阅此 Skill。
Smart Router Skill
Skill by astralwaveorg
blog-writer
为张向阳(Astral Wave)的个人博客 astralwaveorg 生成高质量技术博客文章。当用户说"写博客xxx"、"帮我写一篇xxx"、"引用消息写成博客"、"总结今天聊天",或者 cron 触发每日聊天总结时,使用此技能。文章风格必须模拟作者本人:用工程师第一人称视角,诚实直接,有踩坑说踩坑,读起来像同事之间的经验分享,而不是翻译官方文档。绝不出现"非常优秀""极其强大"等废话词汇。目标是让读者感觉这是技术大牛写的,不是"AI 生成的"。