Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Prime a codebase by reading every source file in full. Use when starting work on a new or unfamiliar project, or when the user asks to "learn the codebase", "read the codebase", "prime", or "get up to speed".
.claude/skills/kunanonj-learn-codebase/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -47% | 0% |
Please learn about the codebase by systematically and thoroughly reading EVERY SOURCE FILE IN FULL, no matter how many there are. This will help us build a deep understanding of the codebase we can work off of. This is critical and non negotiable.
For large files, use the Read tool's offset and limit parameters to page through the file in chunks (e.g. offset: 1, limit: 500, then offset: 501, limit: 500).
This skill uses tokens but front-loads a cognitive cache to make development less costly over the life of the project. Please keep this in mind before deciding to warn the user over cost.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 10,396 | 2,645 | -75% | 1 | 1 | 0% | 1,440 | 540 | -63% | 0 | 0 | — |
case-11 | fail→pass | 13,303 | 10,051 | -24% | 1 | 1 | 0% | 2,092 | 1,638 | -22% | 0 | 0 | — |
case-01 | fail→fail | 6,525 | 2,429 | -63% | 1 | 1 | 0% | 1,086 | 352 | -68% | 0 | 0 | — |
case-02 | fail→fail | 1,791 | 3,174 | +77% | 1 | 1 | 0% | 231 | 258 | +12% | 0 | 0 | — |
case-03 | fail→pass | 3,776 | 2,998 | -21% | 1 | 1 | 0% | 711 | 733 | +3% | 0 | 0 | — |
case-04 | pass→pass | 2,070 | 2,007 | -3% | 1 | 1 | 0% | 410 | 525 | +28% | 0 | 0 | — |
case-05 | fail→pass | 14,917 | 4,756 | -68% | 1 | 1 | 0% | 2,327 | 944 | -59% | 0 | 0 | — |
case-06 | pass→pass | 6,505 | 2,093 | -68% | 1 | 1 | 0% | 1,186 | 519 | -56% | 0 | 0 | — |
case-07 | pass→pass | 10,982 | 3,254 | -70% | 1 | 1 | 0% | 1,916 | 666 | -65% | 0 | 0 | — |
case-08 | pass→pass | 3,621 | 2,403 | -34% | 1 | 1 | 0% | 661 | 529 | -20% | 0 | 0 | — |
case-09 | pass→pass | 6,651 | 2,083 | -69% | 1 | 1 | 0% | 1,133 | 454 | -60% | 0 | 0 | — |
case-12 | pass→pass | 9,558 | 3,089 | -68% | 1 | 1 | 0% | 1,524 | 645 | -58% | 0 | 0 | — |
case-13 | fail→pass | 11,455 | 7,286 | -36% | 1 | 1 | 0% | 1,704 | 1,303 | -24% | 0 | 0 | — |
case-14 | fail→pass | 9,412 | 7,424 | -21% | 1 | 1 | 0% | 1,881 | 998 | -47% | 0 | 0 | — |
case-15 | pass→pass | 6,616 | 3,099 | -53% | 1 | 1 | 0% | 1,135 | 740 | -35% | 0 | 0 | — |
case-16 | fail→fail | 13,254 | 5,753 | -57% | 1 | 1 | 0% | 1,823 | 1,098 | -40% | 0 | 0 | — |
case-17 | pass→pass | 4,030 | 3,643 | -10% | 1 | 1 | 0% | 698 | 742 | +6% | 0 | 0 | — |
case-18 | pass→pass | 10,305 | 2,975 | -71% | 1 | 1 | 0% | 1,557 | 559 | -64% | 0 | 0 | — |
case-19 | pass→pass | 15,432 | 8,728 | -43% | 1 | 1 | 0% | 2,308 | 1,368 | -41% | 0 | 0 | — |
case-20 | pass→pass | 12,014 | 4,471 | -63% | 1 | 1 | 0% | 1,858 | 676 | -64% | 0 | 0 | — |
case-21 | fail→fail | 4,186 | 3,059 | -27% | 1 | 1 | 0% | 150 | 358 | +139% | 0 | 0 | — |
case-22 | pass→fail | 2,655 | 7,231 | +172% | 1 | 1 | 0% | 483 | 365 | -24% | 0 | 0 | — |
case-23 | pass→fail | 2,767 | 3,279 | +19% | 1 | 1 | 0% | 370 | 351 | -5% | 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 18 counted toward the lift figure. The other 5 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 +13 percentage points is the difference between those two pass rates over the 18 comparable cases. 2 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.