Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Task delegation to the heartbeat agent via HEARTBEAT.md. Use when scheduling follow-ups, recurring checks, or recording background jobs for later monitoring.
.claude/skills/x-cmd-claw-heartbeat-delegation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 527% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -31% | 0% |
HEARTBEAT.md is your delegation form to the heartbeat agent. Write clearly what needs to be done and when. The heartbeat agent reads and processes it during idle periods.
## In Progress): After execution, the heartbeat agent marks them completed or removes them.## Recurring): After execution, the heartbeat agent records the run in its own memory/state.yml but keeps the task in HEARTBEAT.md for future runs.> Important: Do NOT execute items in HEARTBEAT.md yourself during the current session. Your role is to read it for awareness of pending follow-ups, but execution is the heartbeat agent's responsibility. If a user asks about a HEARTBEAT.md item, briefly summarize its status; only perform the action if the user explicitly says "do it now" or similar.
Write under ## In Progress:
markdown- **Follow-up**: `<task-id>` - **Due**: YYYY-MM-DD HH:MM - **Task**: <description>
Write under ## Recurring:
markdown- **Recurring**: `<task-id>` - **Frequency**: <hourly | daily | weekly> - **Task**: <description>
When you create a background job via x agent run, also record it in HEARTBEAT.md under ## In Progress:
markdown- **Background Job**: `<job-id>` - **Display Name**: <friendly name> - **IM**: <im> - **Chat ID**: <chatid> - **Task**: <brief description> - **Created**: <timestamp> - **Check command**: `x agent job status --job-id <job-id> --llms` - **Notify command**: `x claw agentrequest <im> <chatid> '<summary>'`
The heartbeat agent periodically checks active background jobs:
x agent job status --job-id "<id>" --llms| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 12,297 | 23,656 | +92% | 1 | 1 | 0% | 814 | 1,028 | +26% | 0 | 0 | — |
case-02 | fail→pass | 5,669 | 6,206 | +9% | 1 | 1 | 0% | 263 | 1,648 | +527% | 0 | 0 | — |
case-03 | fail→fail | 6,277 | 13,746 | +119% | 1 | 1 | 0% | 914 | 851 | -7% | 0 | 0 | — |
case-04 | pass→pass | 10,491 | 8,882 | -15% | 1 | 1 | 0% | 1,599 | 1,815 | +14% | 0 | 0 | — |
case-05 | fail→fail | 6,725 | 11,431 | +70% | 1 | 1 | 0% | 1,043 | 926 | -11% | 0 | 0 | — |
case-06 | pass→pass | 4,616 | 5,069 | +10% | 1 | 1 | 0% | 765 | 1,358 | +78% | 0 | 0 | — |
case-07 | fail→fail | 6,142 | 4,611 | -25% | 1 | 1 | 0% | 151 | 1,050 | +595% | 0 | 0 | — |
case-08 | pass→pass | 4,487 | 7,316 | +63% | 1 | 1 | 0% | 319 | 1,795 | +463% | 0 | 0 | — |
case-09 | fail→fail | 7,099 | 16,715 | +135% | 1 | 1 | 0% | 228 | 2,954 | +1196% | 0 | 0 | — |
case-10 | fail→fail | 5,837 | 16,912 | +190% | 1 | 1 | 0% | 308 | 841 | +173% | 0 | 0 | — |
case-11 | fail→fail | 8,640 | 13,245 | +53% | 1 | 1 | 0% | 757 | 1,203 | +59% | 0 | 0 | — |
case-12 | fail→fail | 5,807 | 12,630 | +117% | 1 | 1 | 0% | 906 | 1,069 | +18% | 0 | 0 | — |
case-13 | fail→pass | 21,868 | 2,609 | -88% | 1 | 1 | 0% | 1,748 | 867 | -50% | 0 | 0 | — |
case-14 | fail→fail | 7,178 | 11,257 | +57% | 1 | 1 | 0% | 230 | 927 | +303% | 0 | 0 | — |
case-15 | fail→pass | 6,448 | 3,366 | -48% | 1 | 1 | 0% | 940 | 1,124 | +20% | 0 | 0 | — |
case-16 | fail→fail | 10,301 | 13,741 | +33% | 1 | 1 | 0% | 503 | 2,003 | +298% | 0 | 0 | — |
case-17 | fail→pass | 7,935 | 2,962 | -63% | 1 | 1 | 0% | 1,162 | 925 | -20% | 0 | 0 | — |
case-18 | fail→pass | 8,368 | 3,275 | -61% | 1 | 1 | 0% | 1,197 | 828 | -31% | 0 | 0 | — |
case-19 | fail→pass | 15,795 | 4,172 | -74% | 1 | 1 | 0% | 2,678 | 1,058 | -60% | 0 | 0 | — |
case-20 | pass→pass | 4,087 | 3,647 | -11% | 1 | 1 | 0% | 536 | 1,074 | +100% | 0 | 0 | — |
case-21 | fail→pass | 9,272 | 5,718 | -38% | 1 | 1 | 0% | 1,305 | 1,529 | +17% | 0 | 0 | — |
case-22 | fail→pass | 7,785 | 10,487 | +35% | 1 | 1 | 0% | 930 | 2,400 | +158% | 0 | 0 | — |
case-23 | fail→fail | 6,982 | 12,311 | +76% | 1 | 1 | 0% | 1,135 | 1,531 | +35% | 0 | 0 | — |
case-24 | fail→pass | 12,081 | 4,049 | -66% | 1 | 1 | 0% | 1,786 | 1,190 | -33% | 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 12 counted toward the lift figure. The other 12 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 +38 percentage points is the difference between those two pass rates over the 12 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.