Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Manage Neon serverless Postgres — projects, branches, databases, roles, and connection strings. Use when an agent needs to create database branches, manage Neon projects, or get connection strings for serverless Postgres.
.claude/skills/composio-community-neon-cli/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 90% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -58% | 0% |
Manage Neon projects, branches, databases, roles, and connection strings.
bashnpm i -g neonctl neonctl auth
bashneonctl projects list neonctl branches list --project-id <id> neonctl branches create --project-id <id> --name feature-branch neonctl databases list --branch <branch-id> neonctl connection-string --project-id <id> neonctl roles list --project-id <id>
--output json for structured outputNEON_API_KEY| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 10,781 | 4,222 | -61% | 1 | 1 | 0% | 625 | 756 | +21% | 0 | 0 | — |
case-02 | fail→pass | 71,457 | 35,778 | -50% | 1 | 1 | 0% | 445 | 844 | +90% | 0 | 0 | — |
case-03 | fail→pass | 10,335 | 63,500 | +514% | 1 | 1 | 0% | 1,481 | 496 | -67% | 0 | 0 | — |
case-04 | fail→pass | 8,347 | 2,444 | -71% | 1 | 1 | 0% | 1,510 | 546 | -64% | 0 | 0 | — |
case-05 | pass→pass | 8,974 | 2,572 | -71% | 1 | 1 | 0% | 1,099 | 551 | -50% | 0 | 0 | — |
case-06 | fail→pass | 7,012 | 3,256 | -54% | 1 | 1 | 0% | 1,274 | 532 | -58% | 0 | 0 | — |
case-11 | fail→pass | 15,563 | 10,425 | -33% | 1 | 1 | 0% | 2,745 | 1,233 | -55% | 0 | 0 | — |
case-07 | fail→pass | 7,151 | 2,416 | -66% | 1 | 1 | 0% | 1,349 | 435 | -68% | 0 | 0 | — |
case-08 | fail→pass | 9,328 | 2,430 | -74% | 1 | 1 | 0% | 1,483 | 442 | -70% | 0 | 0 | — |
case-09 | fail→pass | 8,442 | 2,053 | -76% | 1 | 1 | 0% | 1,388 | 494 | -64% | 0 | 0 | — |
case-10 | fail→pass | 8,327 | 1,703 | -80% | 1 | 1 | 0% | 1,620 | 393 | -76% | 0 | 0 | — |
case-12 | fail→pass | 21,008 | 1,658 | -92% | 1 | 1 | 0% | 1,090 | 377 | -65% | 0 | 0 | — |
case-13 | fail→pass | 10,719 | 1,575 | -85% | 1 | 1 | 0% | 1,541 | 364 | -76% | 0 | 0 | — |
case-14 | fail→pass | 9,030 | 1,253 | -86% | 1 | 1 | 0% | 1,379 | 326 | -76% | 0 | 0 | — |
case-15 | pass→pass | 5,516 | 1,683 | -69% | 1 | 1 | 0% | 1,017 | 449 | -56% | 0 | 0 | — |
case-16 | fail→pass | 15,516 | 1,659 | -89% | 1 | 1 | 0% | 2,696 | 386 | -86% | 0 | 0 | — |
case-17 | fail→pass | 5,233 | 1,830 | -65% | 1 | 1 | 0% | 716 | 430 | -40% | 0 | 0 | — |
case-18 | fail→pass | 13,787 | 1,621 | -88% | 1 | 1 | 0% | 1,880 | 410 | -78% | 0 | 0 | — |
case-19 | pass→pass | 9,900 | 13,503 | +36% | 1 | 1 | 0% | 1,157 | 480 | -59% | 0 | 0 | — |
case-20 | pass→pass | 7,069 | 3,817 | -46% | 1 | 1 | 0% | 892 | 613 | -31% | 0 | 0 | — |
case-21 | pass→pass | 10,021 | 8,020 | -20% | 1 | 1 | 0% | 1,344 | 1,600 | +19% | 0 | 0 | — |
case-22 | pass→pass | 3,769 | 3,719 | -1% | 1 | 1 | 0% | 627 | 811 | +29% | 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 +73 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.