Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Hand the current conversation off to a fresh background agent that picks up the work immediately.
.claude/skills/asymmetric-al-claude-handoff/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 189% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-11 | ✓→✗ | ▼ Worse | 250% | 0% |
Write a handoff summary of the current conversation so a fresh agent can continue the work. Instead of saving it, launch a background agent seeded with the summary as its prompt: claude --bg --name "<descriptive name>" "<handoff summary>". It starts in the current working directory and returns immediately; the user manages it with claude agents.
Always pass -n/--name with a descriptive name (e.g. --name "Fix login bug") — it sets the display name shown in the job list, session picker, and terminal title.
Include a "suggested skills" section in the summary, which suggests skills that the agent should invoke.
Do not duplicate content already captured in other artifacts (PRDs, plans, ADRs, issues, commits, diffs). Reference them by path or URL instead.
Redact any sensitive information, such as API keys, passwords, or personally identifiable information — the summary becomes the agent's prompt.
If the user passed arguments, treat them as a description of what the next session will focus on and tailor the summary accordingly.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,320 | 6,727 | -19% | 1 | 1 | 0% | 1,289 | 715 | -45% | 0 | 0 | — |
case-02 | fail→fail | 10,465 | 6,559 | -37% | 1 | 1 | 0% | 1,652 | 656 | -60% | 0 | 0 | — |
case-03 | fail→fail | 11,252 | 6,185 | -45% | 1 | 1 | 0% | 1,335 | 477 | -64% | 0 | 0 | — |
case-04 | fail→fail | 8,883 | 6,511 | -27% | 1 | 1 | 0% | 1,251 | 661 | -47% | 0 | 0 | — |
case-05 | fail→pass | 4,575 | 4,768 | +4% | 1 | 1 | 0% | 640 | 987 | +54% | 0 | 0 | — |
case-06 | fail→fail | 11,631 | 6,791 | -42% | 1 | 1 | 0% | 2,162 | 726 | -66% | 0 | 0 | — |
case-07 | fail→pass | 7,123 | 14,544 | +104% | 1 | 1 | 0% | 483 | 1,398 | +189% | 0 | 0 | — |
case-08 | pass→pass | 8,974 | 19,417 | +116% | 1 | 1 | 0% | 1,596 | 2,520 | +58% | 0 | 0 | — |
case-09 | pass→pass | 9,737 | 4,481 | -54% | 1 | 1 | 0% | 1,772 | 1,000 | -44% | 0 | 0 | — |
case-10 | fail→pass | 9,009 | 7,386 | -18% | 1 | 1 | 0% | 1,448 | 1,186 | -18% | 0 | 0 | — |
case-11 | pass→fail | 2,411 | 11,337 | +370% | 1 | 1 | 0% | 368 | 1,289 | +250% | 0 | 0 | — |
case-12 | pass→fail | 10,363 | 7,108 | -31% | 1 | 1 | 0% | 2,221 | 650 | -71% | 0 | 0 | — |
case-13 | fail→fail | 8,550 | 16,610 | +94% | 1 | 1 | 0% | 1,287 | 1,091 | -15% | 0 | 0 | — |
case-14 | pass→pass | 8,927 | 4,021 | -55% | 1 | 1 | 0% | 1,346 | 898 | -33% | 0 | 0 | — |
case-15 | fail→fail | 13,031 | 7,479 | -43% | 1 | 1 | 0% | 2,566 | 724 | -72% | 0 | 0 | — |
case-16 | fail→fail | 8,639 | 9,435 | +9% | 1 | 1 | 0% | 1,319 | 609 | -54% | 0 | 0 | — |
case-17 | pass→fail | 9,911 | 13,043 | +32% | 1 | 1 | 0% | 1,667 | 1,061 | -36% | 0 | 0 | — |
case-18 | fail→fail | 10,374 | 13,975 | +35% | 1 | 1 | 0% | 622 | 966 | +55% | 0 | 0 | — |
case-19 | fail→fail | 7,263 | 6,109 | -16% | 1 | 1 | 0% | 1,204 | 434 | -64% | 0 | 0 | — |
case-20 | fail→pass | 11,916 | 4,983 | -58% | 1 | 1 | 0% | 2,247 | 1,163 | -48% | 0 | 0 | — |
case-21 | fail→fail | 7,355 | 11,688 | +59% | 1 | 1 | 0% | 1,285 | 2,502 | +95% | 0 | 0 | — |
case-22 | fail→fail | 5,776 | 6,842 | +18% | 1 | 1 | 0% | 926 | 1,479 | +60% | 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. 22 cases were attempted, and 8 counted toward the lift figure. The other 14 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 +5 percentage points is the difference between those two pass rates over the 8 comparable cases. 3 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.