Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Suggest the matching Convex component when the user hand-rolls a pattern it already solves (crons, sharded-counter, rate-limiter, storage, search, presence, workflow, RAG, prosemirror-sync). Passive — suggest after the task, never interrupt. Never install without consent.
.claude/skills/get-convex-convex-suggest/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -27% | 0% |
<!-- GENERATED from convex-agents content/capabilities/suggest.json — do not edit by hand. -->
When you see code or intent that duplicates what a Convex component already does, surface a targeted suggestion: ONE component, WHY (anchored in the user's own code or ask), and a concrete install hint. Never install without explicit consent. Never suggest more than one component at a time unless the user asks.
/add <component> or follow the installHint from the detector.post.likes + 1 in a mutation that many users call concurrently; that causes OCC conflicts at scale.'| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 17,706 | 9,532 | -46% | 1 | 1 | 0% | 3,469 | 2,234 | -36% | 0 | 0 | — |
case-02 | fail→fail | 13,747 | 8,694 | -37% | 1 | 1 | 0% | 2,626 | 1,851 | -30% | 0 | 0 | — |
case-03 | fail→pass | 13,596 | 10,229 | -25% | 1 | 1 | 0% | 2,479 | 2,290 | -8% | 0 | 0 | — |
case-04 | fail→pass | 23,681 | 8,393 | -65% | 1 | 1 | 0% | 3,346 | 1,709 | -49% | 0 | 0 | — |
case-05 | fail→fail | 11,914 | 10,326 | -13% | 1 | 1 | 0% | 2,264 | 2,228 | -2% | 0 | 0 | — |
case-06 | fail→pass | 24,151 | 14,510 | -40% | 1 | 1 | 0% | 4,427 | 2,862 | -35% | 0 | 0 | — |
case-07 | fail→pass | 17,029 | 10,147 | -40% | 1 | 1 | 0% | 3,466 | 2,064 | -40% | 0 | 0 | — |
case-08 | fail→pass | 16,901 | 9,168 | -46% | 1 | 1 | 0% | 2,841 | 2,073 | -27% | 0 | 0 | — |
case-09 | fail→fail | 15,376 | 11,859 | -23% | 1 | 1 | 0% | 3,214 | 2,286 | -29% | 0 | 0 | — |
case-10 | fail→fail | 21,083 | 10,587 | -50% | 1 | 1 | 0% | 2,938 | 2,557 | -13% | 0 | 0 | — |
case-11 | fail→fail | 23,941 | 11,962 | -50% | 1 | 1 | 0% | 3,500 | 2,858 | -18% | 0 | 0 | — |
case-17 | fail→fail | 14,271 | 9,042 | -37% | 1 | 1 | 0% | 2,646 | 2,207 | -17% | 0 | 0 | — |
case-12 | fail→fail | 14,156 | 11,869 | -16% | 1 | 1 | 0% | 2,702 | 2,352 | -13% | 0 | 0 | — |
case-13 | fail→pass | 25,581 | 9,939 | -61% | 1 | 1 | 0% | 3,927 | 2,204 | -44% | 0 | 0 | — |
case-14 | fail→fail | 22,950 | 14,551 | -37% | 1 | 1 | 0% | 3,283 | 2,799 | -15% | 0 | 0 | — |
case-15 | fail→pass | 21,617 | 11,606 | -46% | 1 | 1 | 0% | 3,400 | 2,572 | -24% | 0 | 0 | — |
case-16 | fail→fail | 19,084 | 12,120 | -36% | 1 | 1 | 0% | 2,855 | 2,116 | -26% | 0 | 0 | — |
case-18 | fail→pass | 8,347 | 6,622 | -21% | 1 | 1 | 0% | 1,520 | 1,755 | +15% | 0 | 0 | — |
case-19 | pass→pass | 9,175 | 4,379 | -52% | 1 | 1 | 0% | 1,661 | 1,396 | -16% | 0 | 0 | — |
case-20 | pass→pass | 6,568 | 5,483 | -17% | 1 | 1 | 0% | 1,276 | 1,635 | +28% | 0 | 0 | — |
case-21 | pass→pass | 13,069 | 8,992 | -31% | 1 | 1 | 0% | 2,369 | 2,350 | -1% | 0 | 0 | — |
case-22 | pass→pass | 10,729 | 7,788 | -27% | 1 | 1 | 0% | 1,473 | 1,880 | +28% | 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. The headline lift of +36 percentage points is the difference between those two pass rates over the 22 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.