Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use ccxt (Python) for crypto prices, balances, and order actions across 100+ exchanges.
.claude/skills/cowork-os-crypto-trading/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 373% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -59% | 0% |
Use ccxt (Python) for crypto prices, balances, and order actions across 100+ exchanges.
| Name | Type | Required | Description | |---|---|---|---| | mode | select | Yes | What operation to run (price, balance, or order). | | exchange | string | Yes | ccxt exchange id (for example binance, kraken, coinbase) | | symbol | string | No | Trading symbol (for example BTC/USDT) | | side | select | No | Order direction for the order action | | order_type | select | No | Order type for the order action | | amount | number | No | Order quantity (base unit, required for order mode) | | price | number | No | Order price per unit (required when order_type is limit) |
../crypto-trading.json.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→fail | 11,149 | 2,593 | -77% | 1 | 1 | 0% | 2,016 | 803 | -60% | 0 | 0 | — |
case-01 | fail→pass | 6,415 | 7,154 | +12% | 1 | 1 | 0% | 1,127 | 1,765 | +57% | 0 | 0 | — |
case-02 | pass→fail | 6,904 | 2,518 | -64% | 1 | 1 | 0% | 1,262 | 627 | -50% | 0 | 0 | — |
case-03 | pass→pass | 8,185 | 6,846 | -16% | 1 | 1 | 0% | 1,433 | 1,695 | +18% | 0 | 0 | — |
case-04 | pass→pass | 9,757 | 7,116 | -27% | 1 | 1 | 0% | 1,729 | 1,755 | +2% | 0 | 0 | — |
case-05 | pass→pass | 5,036 | 5,528 | +10% | 1 | 1 | 0% | 824 | 1,443 | +75% | 0 | 0 | — |
case-07 | fail→pass | 7,999 | 7,596 | -5% | 1 | 1 | 0% | 1,379 | 1,662 | +21% | 0 | 0 | — |
case-08 | pass→pass | 5,987 | 5,131 | -14% | 1 | 1 | 0% | 1,020 | 1,314 | +29% | 0 | 0 | — |
case-09 | pass→pass | 3,938 | 5,190 | +32% | 1 | 1 | 0% | 619 | 1,313 | +112% | 0 | 0 | — |
case-10 | pass→pass | 10,497 | 4,634 | -56% | 1 | 1 | 0% | 1,974 | 1,153 | -42% | 0 | 0 | — |
case-11 | fail→fail | 5,494 | 2,190 | -60% | 1 | 1 | 0% | 841 | 766 | -9% | 0 | 0 | — |
case-12 | pass→pass | 5,575 | 2,848 | -49% | 1 | 1 | 0% | 975 | 930 | -5% | 0 | 0 | — |
case-13 | fail→pass | 3,892 | 12,927 | +232% | 1 | 1 | 0% | 551 | 2,604 | +373% | 0 | 0 | — |
case-14 | fail→fail | 5,973 | 5,015 | -16% | 1 | 1 | 0% | 1,092 | 729 | -33% | 0 | 0 | — |
case-15 | pass→pass | 5,391 | 4,265 | -21% | 1 | 1 | 0% | 957 | 1,136 | +19% | 0 | 0 | — |
case-16 | pass→fail | 3,169 | 2,625 | -17% | 1 | 1 | 0% | 408 | 767 | +88% | 0 | 0 | — |
case-17 | pass→pass | 10,674 | 3,714 | -65% | 1 | 1 | 0% | 1,702 | 969 | -43% | 0 | 0 | — |
case-18 | fail→pass | 4,872 | 7,079 | +45% | 1 | 1 | 0% | 646 | 1,111 | +72% | 0 | 0 | — |
case-19 | fail→pass | 8,113 | 1,528 | -81% | 1 | 1 | 0% | 1,422 | 590 | -59% | 0 | 0 | — |
case-20 | pass→pass | 16,661 | 9,375 | -44% | 1 | 1 | 0% | 2,860 | 2,066 | -28% | 0 | 0 | — |
case-21 | pass→pass | 17,818 | 7,171 | -60% | 1 | 1 | 0% | 2,708 | 1,469 | -46% | 0 | 0 | — |
case-22 | pass→pass | 17,408 | 11,227 | -36% | 1 | 1 | 0% | 2,977 | 2,273 | -24% | 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 +9 percentage points is the difference between those two pass rates over the 22 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.