Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Unit-aware computation with Pint - convert units, dimensional analysis, unit arithmetic
.claude/skills/pint-compute/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 134% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 5% | 0% |
Cognitive prosthetics for unit-aware computation. Use Pint for converting between units, performing unit arithmetic, checking dimensional compatibility, and simplifying compound units.
| I want to... | Command | Example | |--------------|---------|---------| | Convert units | convert | convert "5 meters" --to feet | | Unit math | calc | calc "10 m/s * 5 s" | | Check dimensions | check | check newton --against "kg * m / s^2" | | Parse quantity | parse | parse "100 km/h" | | Simplify units | simplify | simplify "1 kg*m/s^2" |
Parse a quantity string into magnitude, units, and dimensionality.
bashuv run python -m runtime.harness scripts/pint_compute.py \ parse "100 km/h" uv run python -m runtime.harness scripts/pint_compute.py \ parse "9.8 m/s^2"
Convert a quantity to different units.
bashuv run python -m runtime.harness scripts/pint_compute.py \ convert "5 meters" --to feet uv run python -m runtime.harness scripts/pint_compute.py \ convert "100 km/h" --to mph uv run python -m runtime.harness scripts/pint_compute.py \ convert "1 atmosphere" --to pascal
Perform unit-aware arithmetic. Operators must be space-separated.
bashuv run python -m runtime.harness scripts/pint_compute.py \ calc "5 m * 3 s" uv run python -m runtime.harness scripts/pint_compute.py \ calc "10 m / 2 s" uv run python -m runtime.harness scripts/pint_compute.py \ calc "5 meters + 300 cm"
Check if two units have compatible dimensions.
bashuv run python -m runtime.harness scripts/pint_compute.py \ check newton --against "kg * m / s^2" uv run python -m runtime.harness scripts/pint_compute.py \ check joule --against "kg * m^2 / s^2"
Simplify compound units to base or compact form.
bashuv run python -m runtime.harness scripts/pint_compute.py \ simplify "1 kg*m/s^2" uv run python -m runtime.harness scripts/pint_compute.py \ simplify "1000 m"
| Domain | Examples | |--------|----------| | Length | meter, foot, inch, mile, km, yard | | Time | second, minute, hour, day, year | | Mass | kg, gram, pound, ounce, ton | | Velocity | m/s, km/h, mph, knot | | Energy | joule, calorie, eV, kWh, BTU | | Force | newton, pound_force, dyne | | Temperature | kelvin, celsius, fahrenheit | | Pressure | pascal, bar, atmosphere, psi | | Power | watt, horsepower |
All commands return JSON with relevant fields:
json{ "result": "16.4042 foot", "magnitude": 16.4042, "units": "foot", "dimensionality": "[length]", "latex": "16.4042\\,\\mathrm{ft}" }
Dimensionality errors are caught and reported:
bash# This will error - incompatible dimensions uv run python -m runtime.harness scripts/pint_compute.py \ convert "5 meters" --to kg # Error: Cannot convert '[length]' to '[mass]'
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | fail→pass | 2,729 | 3,379 | +24% | 1 | 1 | 0% | 494 | 1,155 | +134% | 0 | 0 | — |
case-01 | fail→fail | 4,987 | 4,628 | -7% | 1 | 1 | 0% | 1,005 | 1,198 | +19% | 0 | 0 | — |
case-02 | fail→fail | 3,346 | 4,344 | +30% | 1 | 1 | 0% | 520 | 1,205 | +132% | 0 | 0 | — |
case-03 | fail→fail | 8,459 | 4,307 | -49% | 1 | 1 | 0% | 1,726 | 1,223 | -29% | 0 | 0 | — |
case-04 | fail→pass | 7,153 | 1,346 | -81% | 1 | 1 | 0% | 1,075 | 1,189 | +11% | 0 | 0 | — |
case-05 | fail→fail | 8,477 | 4,240 | -50% | 1 | 1 | 0% | 1,642 | 1,170 | -29% | 0 | 0 | — |
case-06 | fail→pass | 5,308 | 2,992 | -44% | 1 | 1 | 0% | 959 | 1,144 | +19% | 0 | 0 | — |
case-07 | fail→pass | 5,789 | 1,921 | -67% | 1 | 1 | 0% | 1,102 | 1,141 | +4% | 0 | 0 | — |
case-08 | fail→fail | 5,386 | 3,983 | -26% | 1 | 1 | 0% | 1,035 | 1,162 | +12% | 0 | 0 | — |
case-09 | fail→fail | 5,449 | 3,917 | -28% | 1 | 1 | 0% | 901 | 1,166 | +29% | 0 | 0 | — |
case-11 | fail→pass | 5,545 | 1,808 | -67% | 1 | 1 | 0% | 1,130 | 1,183 | +5% | 0 | 0 | — |
case-12 | fail→fail | 2,949 | 4,794 | +63% | 1 | 1 | 0% | 481 | 1,159 | +141% | 0 | 0 | — |
case-13 | fail→fail | 4,634 | 4,277 | -8% | 1 | 1 | 0% | 936 | 1,161 | +24% | 0 | 0 | — |
case-14 | fail→fail | 4,330 | 5,757 | +33% | 1 | 1 | 0% | 776 | 1,384 | +78% | 0 | 0 | — |
case-15 | fail→fail | 3,117 | 4,747 | +52% | 1 | 1 | 0% | 547 | 1,171 | +114% | 0 | 0 | — |
case-16 | fail→fail | 3,507 | 4,515 | +29% | 1 | 1 | 0% | 609 | 1,190 | +95% | 0 | 0 | — |
case-17 | fail→fail | 4,276 | 1,404 | -67% | 1 | 1 | 0% | 742 | 1,177 | +59% | 0 | 0 | — |
case-18 | fail→pass | 6,088 | 1,574 | -74% | 1 | 1 | 0% | 1,104 | 1,246 | +13% | 0 | 0 | — |
case-19 | fail→pass | 5,140 | 1,852 | -64% | 1 | 1 | 0% | 1,043 | 1,164 | +12% | 0 | 0 | — |
case-20 | fail→pass | 6,906 | 1,295 | -81% | 1 | 1 | 0% | 1,276 | 1,197 | -6% | 0 | 0 | — |
case-21 | pass→pass | 4,735 | 4,740 | +0% | 1 | 1 | 0% | 777 | 1,960 | +152% | 0 | 0 | — |
case-22 | pass→pass | 8,612 | 8,876 | +3% | 1 | 1 | 0% | 1,724 | 2,704 | +57% | 0 | 0 | — |
case-23 | pass→fail | 2,258 | 1,919 | -15% | 1 | 1 | 0% | 441 | 1,273 | +189% | 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. 23 cases were attempted, and 12 counted toward the lift figure. The other 11 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 -100 percentage points is the difference between those two pass rates over the 12 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 7/29/2026 | — |
Other measured skills in the registry, with their headline benchmark lift.