Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Nordic Semiconductor nRF5x/nRF Connect SDK expertise
.claude/skills/a5c-ai-nordic-nrf/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-21 | ✓→✗ | ▼ Worse | -2% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -26% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 35% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -22% | 0% |
This skill provides expert-level support for Nordic Semiconductor nRF5x series and nRF Connect SDK development, with deep expertise in Bluetooth Low Energy, power optimization, and wireless protocols.
bsp-development.js - nRF BSP implementationlow-power-design.js - Ultra-low-power nRF designpower-consumption-profiling.js - Power analysisota-firmware-update.js - nRF DFU implementationThis skill is invoked when tasks require:
| Series | Features | |--------|----------| | nRF51 | Legacy BLE (deprecated) | | nRF52810/832/833 | BLE 5.0, low-cost | | nRF52820 | BLE 5.2, Thread | | nRF52840 | BLE 5.0, Thread, Zigbee, USB | | nRF5340 | Dual-core, BLE 5.2 | | nRF9160 | LTE-M/NB-IoT cellular |
cstatic struct bt_le_adv_param adv_param = BT_LE_ADV_PARAM_INIT( BT_LE_ADV_OPT_CONNECTABLE | BT_LE_ADV_OPT_USE_NAME, BT_GAP_ADV_FAST_INT_MIN_2, BT_GAP_ADV_FAST_INT_MAX_2, NULL); bt_le_adv_start(&adv_param, ad, ARRAY_SIZE(ad), NULL, 0);
kconfigCONFIG_PM=y CONFIG_PM_DEVICE=y CONFIG_BT_CTLR_TX_PWR_MINUS_8=y CONFIG_BT_CTLR_ADV_EXT=n
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→pass | 11,552 | 5,849 | -49% | 1 | 1 | 0% | 2,094 | 1,556 | -26% | 0 | 0 | — |
case-01 | fail→pass | 24,373 | 20,595 | -16% | 1 | 1 | 0% | 3,783 | 3,321 | -12% | 0 | 0 | — |
case-02 | fail→fail | 24,751 | 20,788 | -16% | 1 | 1 | 0% | 4,103 | 4,559 | +11% | 0 | 0 | — |
case-03 | pass→pass | 9,637 | 10,928 | +13% | 1 | 1 | 0% | 1,757 | 2,367 | +35% | 0 | 0 | — |
case-04 | pass→pass | 11,397 | 4,834 | -58% | 1 | 1 | 0% | 1,985 | 1,543 | -22% | 0 | 0 | — |
case-06 | pass→pass | 6,378 | 3,618 | -43% | 1 | 1 | 0% | 884 | 1,333 | +51% | 0 | 0 | — |
case-07 | pass→pass | 6,832 | 6,597 | -3% | 1 | 1 | 0% | 979 | 1,726 | +76% | 0 | 0 | — |
case-08 | pass→pass | 4,862 | 3,502 | -28% | 1 | 1 | 0% | 814 | 1,324 | +63% | 0 | 0 | — |
case-09 | pass→pass | 5,233 | 4,689 | -10% | 1 | 1 | 0% | 666 | 1,478 | +122% | 0 | 0 | — |
case-10 | pass→pass | 13,709 | 11,384 | -17% | 1 | 1 | 0% | 2,066 | 2,443 | +18% | 0 | 0 | — |
case-11 | pass→pass | 5,610 | 5,132 | -9% | 1 | 1 | 0% | 887 | 1,452 | +64% | 0 | 0 | — |
case-12 | pass→pass | 6,168 | 5,067 | -18% | 1 | 1 | 0% | 1,052 | 1,456 | +38% | 0 | 0 | — |
case-13 | pass→pass | 20,481 | 13,229 | -35% | 1 | 1 | 0% | 3,280 | 2,899 | -12% | 0 | 0 | — |
case-14 | pass→pass | 15,430 | 12,314 | -20% | 1 | 1 | 0% | 2,515 | 3,084 | +23% | 0 | 0 | — |
case-15 | pass→pass | 5,802 | 6,058 | +4% | 1 | 1 | 0% | 976 | 1,770 | +81% | 0 | 0 | — |
case-16 | pass→pass | 2,552 | 2,639 | +3% | 1 | 1 | 0% | 438 | 1,188 | +171% | 0 | 0 | — |
case-17 | pass→pass | 13,947 | 12,941 | -7% | 1 | 1 | 0% | 1,881 | 2,504 | +33% | 0 | 0 | — |
case-18 | pass→pass | 11,218 | 10,501 | -6% | 1 | 1 | 0% | 1,852 | 2,467 | +33% | 0 | 0 | — |
case-19 | pass→pass | 7,238 | 6,641 | -8% | 1 | 1 | 0% | 924 | 1,788 | +94% | 0 | 0 | — |
case-20 | pass→pass | 16,746 | 15,770 | -6% | 1 | 1 | 0% | 2,727 | 3,457 | +27% | 0 | 0 | — |
case-21 | pass→fail | 24,218 | 18,628 | -23% | 1 | 1 | 0% | 4,031 | 3,943 | -2% | 0 | 0 | — |
case-22 | pass→pass | 14,541 | 11,485 | -21% | 1 | 1 | 0% | 2,441 | 3,200 | +31% | 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 0 percentage points is the difference between those two pass rates over the 22 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.
Other measured skills in the registry, with their headline benchmark lift.