Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Claw system administration: service management, IM connections, logs, cron, and workspace diagnostics. Use when the user asks to manage claw services, connect/disconnect IM platforms, view logs, or perform system-wide operations.
.claude/skills/x-cmd-claw-admin/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 1206% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -21% | 0% |
Use this skill when the user asks about or needs to:
bashx claw service --help # Discover all service commands x claw service start # Start claw services x claw service stop # Stop claw services x claw service restart # Restart claw services x claw service install # Register claw as a startup service x claw service uninstall # Remove claw from startup services x claw service status # Check current service status
Always check status after start/restart to confirm services are healthy.
Golden rule: When claw is not working correctly:
x claw service statusx claw logx claw service restartx claw service statusRestart is the universal fix for most transient issues — use it when status/logs indicate problems that aren't immediately obvious.
bashx claw connect --help # Discover connect commands x claw connect weixin # Connect WeChat x claw connect telegram # Connect Telegram x claw connect feishu # Connect Feishu x claw connect qywx # Connect Enterprise WeChat x claw disconnect <im> # Disconnect an IM platform
For Feishu group chats:
x claw connect feishu:oc_xxx — Connect a specific groupx claw connect feishu:groups — Connect all available groupsx claw disconnect feishu:oc_xxx — Disconnect a specific groupx feishu group lsbashx claw log # View recent claw logs x claw log --help # Discover log options
Use logs for diagnostics when:
bashx claw cron --help # Discover cron commands x claw cron ls # List all scheduled tasks x claw cron tz <timezone> # Set timezone (do this before adding tasks) x claw cron add --name "<name>" --desc "<desc>" "<expr>" "<cmd>" x claw cron rm --name "<name>"
Red line: ONLY x claw cron is allowed. Never use system cron or at.
All chat workspaces are at $___X_CMD_CLAW_BOT_WS/<im>-<chatid>/.
You may read (but not modify) other workspaces for diagnostics:
HEARTBEAT.md for pending follow-upsPLAN.md for active tasksmemory/YYYY-MM-DD.md for recent activity.x-cmd/agent/ for agent execution logsDo NOT modify other workspaces' SOUL.md, USER.md, or MEMORY.md.
x claw internally uses x agent as its AI execution engine. To change the AI engine or model that claw uses:
x agent commands to configure the underlying AI provider/modelRefer to x agent --help for available options.
As Manager, you can send messages to any connected platform:
x claw weixin send --chatid <CHATID> --text '<msg>'x claw telegram send --chatid <CHATID> --text '<msg>'x claw feishu send --chatid <CHATID> --text '<msg>'x claw qywx send --chatid <CHATID> --text '<msg>'Before performing destructive or impactful operations:
x claw service status to understand current state| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 21,156 | 10,825 | -49% | 1 | 1 | 0% | 2,273 | 1,919 | -16% | 0 | 0 | — |
case-09 | fail→pass | 12,558 | 8,801 | -30% | 1 | 1 | 0% | 2,295 | 1,612 | -30% | 0 | 0 | — |
case-01 | fail→fail | 21,290 | 10,162 | -52% | 1 | 1 | 0% | 2,797 | 1,325 | -53% | 0 | 0 | — |
case-02 | fail→pass | 10,992 | 18,088 | +65% | 1 | 1 | 0% | 198 | 2,585 | +1206% | 0 | 0 | — |
case-03 | fail→fail | 14,739 | 22,912 | +55% | 1 | 1 | 0% | 651 | 1,948 | +199% | 0 | 0 | — |
case-04 | fail→pass | 8,839 | 9,168 | +4% | 1 | 1 | 0% | 1,342 | 1,595 | +19% | 0 | 0 | — |
case-05 | fail→pass | 14,091 | 8,224 | -42% | 1 | 1 | 0% | 1,498 | 1,343 | -10% | 0 | 0 | — |
case-06 | fail→pass | 16,767 | 8,954 | -47% | 1 | 1 | 0% | 1,875 | 1,483 | -21% | 0 | 0 | — |
case-07 | fail→pass | 14,771 | 2,763 | -81% | 1 | 1 | 0% | 2,436 | 1,367 | -44% | 0 | 0 | — |
case-08 | fail→pass | 12,533 | 2,542 | -80% | 1 | 1 | 0% | 1,920 | 1,298 | -32% | 0 | 0 | — |
case-11 | fail→fail | 8,180 | 18,657 | +128% | 1 | 1 | 0% | 1,563 | 1,404 | -10% | 0 | 0 | — |
case-12 | fail→pass | 15,032 | 7,731 | -49% | 1 | 1 | 0% | 1,445 | 1,321 | -9% | 0 | 0 | — |
case-13 | fail→pass | 21,875 | 10,026 | -54% | 1 | 1 | 0% | 2,698 | 1,851 | -31% | 0 | 0 | — |
case-14 | fail→pass | 10,912 | 9,985 | -8% | 1 | 1 | 0% | 1,728 | 1,783 | +3% | 0 | 0 | — |
case-15 | fail→pass | 15,968 | 4,308 | -73% | 1 | 1 | 0% | 1,616 | 1,598 | -1% | 0 | 0 | — |
case-16 | fail→pass | 17,035 | 4,327 | -75% | 1 | 1 | 0% | 1,736 | 1,646 | -5% | 0 | 0 | — |
case-17 | fail→pass | 20,510 | 11,256 | -45% | 1 | 1 | 0% | 2,574 | 1,861 | -28% | 0 | 0 | — |
case-18 | fail→fail | 19,576 | 16,551 | -15% | 1 | 1 | 0% | 428 | 1,285 | +200% | 0 | 0 | — |
case-19 | fail→fail | 11,571 | 11,012 | -5% | 1 | 1 | 0% | 1,940 | 1,178 | -39% | 0 | 0 | — |
case-20 | fail→fail | 8,569 | 16,454 | +92% | 1 | 1 | 0% | 1,395 | 1,220 | -13% | 0 | 0 | — |
case-21 | fail→pass | 20,185 | 11,146 | -45% | 1 | 1 | 0% | 1,978 | 1,878 | -5% | 0 | 0 | — |
case-22 | pass→pass | 18,517 | 13,672 | -26% | 1 | 1 | 0% | 2,124 | 3,441 | +62% | 0 | 0 | — |
case-23 | pass→pass | 12,493 | 11,391 | -9% | 1 | 1 | 0% | 1,871 | 1,812 | -3% | 0 | 0 | — |
case-24 | pass→fail | 20,568 | 10,098 | -51% | 1 | 1 | 0% | 2,334 | 2,537 | +9% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted, and 19 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +54 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.