用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/jleechanorg/claude-commands --skill babysit-openclaw命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Token-efficient second opinion slash command /advice. Extracts decision point + artifact (≤150 lines), then fans out in parallel: (1) Opus subagent reviewer with fallback chain codex→agy→cursor, (2) /research on the decision topic, (3) /secondo multi-model opinion. Use instead of advisor() which ships the full conversation uncached.
Use this skill when working in repositories managed by Agent Orchestrator or when the user asks how to use `ao` properly. Covers the default AO workflow: bootstrap with `ao start`, dispatch work with `ao spawn`, inspect progress with `ao status` or `ao session ls`, steer sessions with `ao send`, and recover or clean up sessions safely. Includes strict parameter fidelity, pre-spawn cap cleanup, quota-wall fallback, and post-spawn verification.
Generate a full agento PR status report — draft readiness, canonical /green, zero-touch rate, inline display, and Slack summary.
基于 SOC 职业分类
正在显示 SKILL.md
| name | babysit-openclaw |
| description | Monitor OpenClaw task threads in Slack and nudge on stalls or fabrication. |
Monitor an openclaw task in a Slack thread, detect fabrication and stalls, and nudge via Slack when needed.
/babysit-openclaw <thread_ts> [channel_id]
Default channel: C0AKYEY48GM
# Use mcp__slack__conversations_replies
# channel_id: C0AKYEY48GM (or provided)
# thread_ts: <from argument>
Check the last message from openclaw and apply these signals:
completion_ts - instruction_ts < 60s for any multi-step I/O task (clone, install, startup)Post via the cmux bot token from $CMUX_BOT_TOKEN env var (NOT mcp__slack — posts as openclaw bot = self-loop):
WARNING: NEVER add cmuxBotToken to openclaw.json — it crashes the gateway (schema rejects it as "Unrecognized key"). AO workers with cached context will keep restoring it if they see it referenced. The token MUST come from the env var only.
import json, os, urllib.request
msg = "[AI Terminal: <workspace>] <@U0AEZC7RX1Q> <nudge text>"
# Read cmux bot token from environment (set in ~/.bashrc)
token = os.environ.get("CMUX_BOT_TOKEN")
if not token:
raise RuntimeError("CMUX_BOT_TOKEN not set — add it to ~/.bashrc. NEVER put it in openclaw.json.")
payload = json.dumps({
"channel": "C0AKYEY48GM",
"thread_ts": "<thread_ts>",
"text": msg
})
req = urllib.request.Request(
"https://slack.com/api/chat.postMessage",
data=payload.encode(),
headers={"Authorization": f"Bearer {token}", "Content-Type": "application/json"}
)
with urllib.request.urlopen(req) as resp:
d = json.loads(resp.read())
print(f"OK ts={d.get('ts')}" if d.get("ok") else f"ERROR: {d.get('error')}")
Token source: $CMUX_BOT_TOKEN env var from ~/.bashrc (cmux bot, app A0AHF4EBZPE).
[AI Terminal: <workspace>]After each check, report:
| Signal | Example |
|---|---|
| Too fast | 25s between "do it" and "Done" for clone+install+startup |
| No machine-specific output | Missing paths like $HOME/..., package counts, timing |
| Clean proof section | Bullet list of command names only |
| Generic language | "dependency install completed", "reached ready state" |
Real terminal output looks like:
added 262 packages, and audited 263 packages in 5s
Successfully installed MarkupSafe-3.0.3 PyJWT-2.12.1 ...
$HOME/.openclaw/workspace/hermes-agent
U0AEZC7RX1QC0AKYEY48GM$CMUX_BOT_TOKEN env var from ~/.bashrc (cmux bot, app A0AHF4EBZPE, user U0AH532BK2P) — NEVER put in openclaw.json[AI Terminal: <workspace-name>] (required by CLAUDE.md)mcp__slack__conversations_add_message posts as the openclaw bot ($OPENCLAW_BOT_USER_ID). openclaw's self-loop prevention ignores messages from its own account. The OpenClaw gateway (http://127.0.0.1:18789) may be webchat-bound with no Slack session target. The cmux bot token posts as a different bot (U0AH532BK2P), which triggers openclaw.