Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Merge the winning agent's branch into base, archive losers, and clean up worktrees. Use when the user runs /hub:merge or asks to land the winning AgentHub result and tidy the session.
.claude/skills/alirezarezvani-merge/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 1585% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -15% | 0% |
Merge the best agent's branch into the base branch, archive losing branches via git tags, and clean up worktrees.
/hub:merge # Merge winner of latest session
/hub:merge 20260317-143022 # Merge winner of specific session
/hub:merge 20260317-143022 --agent agent-2 # Explicitly choose winnerIf --agent specified, use that. Otherwise, use the #1 ranked agent from the most recent /hub:eval.
bashgit checkout {base_branch} git merge --no-ff hub/{session-id}/{winner}/attempt-1 \ -m "hub: merge {winner} from session {session-id} Task: {task} Winner: {winner} Session: {session-id}"
For each non-winning agent:
bash# Create archive tag (preserves commits forever) git tag hub/archive/{session-id}/{agent-id} hub/{session-id}/{agent-id}/attempt-1 # Delete branch ref (commits preserved via tag) git branch -D hub/{session-id}/{agent-id}/attempt-1
bashpython {skill_path}/scripts/session_manager.py --cleanup {session-id}
Write .agenthub/board/results/merge-summary.md:
markdown--- author: coordinator timestamp: {now} channel: results --- ## Merge Summary - **Session**: {session-id} - **Winner**: {winner} - **Merged into**: {base_branch} - **Archived**: {loser-1}, {loser-2}, ... - **Worktrees cleaned**: {count}
bashpython {skill_path}/scripts/session_manager.py --update {session-id} --state merged
--no-ff for clear historyTell the user:
{base_branch}hub/archive/{session-id}/agent-{N}merged| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | 12,152 | 5,149 | -58% | 1 | 1 | 0% | 2,546 | 1,045 | -59% | 0 | 0 | — |
case-01 | fail→fail | 2,815 | 5,150 | +83% | 1 | 1 | 0% | 358 | 940 | +163% | 0 | 0 | — |
case-02 | fail→fail | 6,351 | 5,428 | -15% | 1 | 1 | 0% | 1,323 | 928 | -30% | 0 | 0 | — |
case-04 | fail→pass | 3,776 | 15,026 | +298% | 1 | 1 | 0% | 238 | 4,010 | +1585% | 0 | 0 | — |
case-05 | pass→fail | 1,657 | 3,828 | +131% | 1 | 1 | 0% | 275 | 946 | +244% | 0 | 0 | — |
case-06 | fail→fail | 5,629 | 3,997 | -29% | 1 | 1 | 0% | 398 | 825 | +107% | 0 | 0 | — |
case-07 | fail→pass | 8,344 | 3,705 | -56% | 1 | 1 | 0% | 1,466 | 1,335 | -9% | 0 | 0 | — |
case-08 | fail→pass | 9,580 | 3,125 | -67% | 1 | 1 | 0% | 1,747 | 1,278 | -27% | 0 | 0 | — |
case-09 | fail→pass | 6,792 | 1,973 | -71% | 1 | 1 | 0% | 1,435 | 1,002 | -30% | 0 | 0 | — |
case-10 | fail→pass | 5,776 | 1,906 | -67% | 1 | 1 | 0% | 1,214 | 1,030 | -15% | 0 | 0 | — |
case-11 | pass→pass | 5,567 | 2,667 | -52% | 1 | 1 | 0% | 1,162 | 1,186 | +2% | 0 | 0 | — |
case-12 | fail→pass | 9,086 | 2,509 | -72% | 1 | 1 | 0% | 1,766 | 1,154 | -35% | 0 | 0 | — |
case-13 | fail→pass | 10,119 | 1,431 | -86% | 1 | 1 | 0% | 1,953 | 783 | -60% | 0 | 0 | — |
case-14 | fail→pass | 9,622 | 1,379 | -86% | 1 | 1 | 0% | 1,699 | 800 | -53% | 0 | 0 | — |
case-15 | fail→pass | 6,827 | 1,521 | -78% | 1 | 1 | 0% | 1,249 | 856 | -31% | 0 | 0 | — |
case-16 | fail→pass | 6,460 | 3,777 | -42% | 1 | 1 | 0% | 1,356 | 1,421 | +5% | 0 | 0 | — |
case-17 | pass→pass | 8,494 | 5,310 | -37% | 1 | 1 | 0% | 1,543 | 1,683 | +9% | 0 | 0 | — |
case-18 | fail→pass | 8,860 | 1,192 | -87% | 1 | 1 | 0% | 1,465 | 813 | -45% | 0 | 0 | — |
case-19 | pass→pass | 2,621 | 2,089 | -20% | 1 | 1 | 0% | 406 | 1,070 | +164% | 0 | 0 | — |
case-20 | pass→pass | 11,397 | 2,480 | -78% | 1 | 1 | 0% | 1,975 | 1,104 | -44% | 0 | 0 | — |
case-21 | fail→pass | 9,244 | 3,062 | -67% | 1 | 1 | 0% | 1,698 | 1,218 | -28% | 0 | 0 | — |
case-22 | pass→pass | 5,023 | 2,014 | -60% | 1 | 1 | 0% | 1,019 | 1,035 | +2% | 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 16 counted toward the lift figure. The other 6 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 +50 percentage points is the difference between those two pass rates over the 16 comparable cases. 1 case got worse with the skill loaded, and it is 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.