Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Power-user audit of Origin's pending surfaces. Most users want `/brief` for revisions. That handles the daily flow. Use `/review` only for explicit deep-walk audits after bulk imports, or when you want to walk the full queue rather than the top 3 shown in /brief. Invoked as `/review captures` or `/review revisions`.
.claude/skills/davepoon-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -53% | 0% |
Power-user audit lever. Most users do not need /review in daily flow:
/brief automatically (top 3 with inline accept/dismiss)./handoff's previewblock (top 3, informational). Use /review captures for the deep walk.
/distill's topic-suggestion block.Use /review only when you want the deep walk those skills intentionally do not force.
/review captures: walk every unconfirmed memory (list_pending,unfiltered by session). Per item: accept (confirm_memory), edit (capture with supersedes=<old_id> then forget(old_id)), or reject (forget).
/review revisions: walk every pending revision (list_pending_revisions,no cap). Per item: accept (accept_revision), dismiss (dismiss_revision), or skip.
Bare /review (no arg) prints this help block and exits. Does not auto-walk.
every auto-classification before sealing.
/brief shows ">3 pending revisions" and you want to clear thefull queue, not just the top 3.
/brief handles the surface that matters today./recall./recall.Read-only until the user confirms or rejects. No LLM calls. Cheap.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,299 | 3,723 | -30% | 1 | 1 | 0% | 723 | 565 | -22% | 0 | 0 | — |
case-02 | fail→fail | 5,094 | 3,345 | -34% | 1 | 1 | 0% | 744 | 519 | -30% | 0 | 0 | — |
case-03 | fail→pass | 6,645 | 3,563 | -46% | 1 | 1 | 0% | 1,084 | 999 | -8% | 0 | 0 | — |
case-04 | fail→pass | 6,977 | 3,733 | -46% | 1 | 1 | 0% | 1,280 | 979 | -24% | 0 | 0 | — |
case-05 | fail→pass | 7,724 | 2,084 | -73% | 1 | 1 | 0% | 1,379 | 713 | -48% | 0 | 0 | — |
case-06 | fail→pass | 4,947 | 1,699 | -66% | 1 | 1 | 0% | 797 | 636 | -20% | 0 | 0 | — |
case-07 | pass→pass | 2,920 | 1,806 | -38% | 1 | 1 | 0% | 522 | 604 | +16% | 0 | 0 | — |
case-08 | fail→pass | 7,416 | 1,672 | -77% | 1 | 1 | 0% | 1,376 | 648 | -53% | 0 | 0 | — |
case-09 | fail→pass | 7,687 | 3,038 | -60% | 1 | 1 | 0% | 1,393 | 936 | -33% | 0 | 0 | — |
case-10 | fail→pass | 7,436 | 2,350 | -68% | 1 | 1 | 0% | 1,192 | 774 | -35% | 0 | 0 | — |
case-11 | fail→pass | 13,439 | 2,175 | -84% | 1 | 1 | 0% | 2,460 | 722 | -71% | 0 | 0 | — |
case-12 | fail→pass | 2,441 | 2,663 | +9% | 1 | 1 | 0% | 362 | 704 | +94% | 0 | 0 | — |
case-13 | fail→pass | 10,659 | 2,389 | -78% | 1 | 1 | 0% | 1,859 | 754 | -59% | 0 | 0 | — |
case-14 | fail→pass | 10,550 | 2,767 | -74% | 1 | 1 | 0% | 1,849 | 826 | -55% | 0 | 0 | — |
case-15 | fail→pass | 11,059 | 2,859 | -74% | 1 | 1 | 0% | 1,801 | 779 | -57% | 0 | 0 | — |
case-16 | fail→pass | 9,610 | 1,983 | -79% | 1 | 1 | 0% | 1,767 | 682 | -61% | 0 | 0 | — |
case-17 | pass→pass | 7,236 | 2,203 | -70% | 1 | 1 | 0% | 1,129 | 718 | -36% | 0 | 0 | — |
case-18 | pass→pass | 8,147 | 2,482 | -70% | 1 | 1 | 0% | 1,202 | 767 | -36% | 0 | 0 | — |
case-19 | fail→pass | 8,819 | 2,002 | -77% | 1 | 1 | 0% | 1,309 | 722 | -45% | 0 | 0 | — |
case-20 | fail→pass | 7,386 | 1,975 | -73% | 1 | 1 | 0% | 1,118 | 723 | -35% | 0 | 0 | — |
case-21 | fail→pass | 23,543 | 1,613 | -93% | 1 | 1 | 0% | 3,969 | 613 | -85% | 0 | 0 | — |
case-22 | fail→pass | 6,370 | 3,084 | -52% | 1 | 1 | 0% | 1,087 | 880 | -19% | 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 20 counted toward the lift figure. The other 2 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 +77 percentage points is the difference between those two pass rates over the 20 comparable cases.
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.