Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Presents adaptive codebase challenge questions with multiple-choice and trace exercises. Use when testing contributor knowledge of the codebase.
.claude/skills/athola-challenge/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -36% | 0% |
Present challenges from the knowledge base and evaluate answers.
gauntlet:extract)gauntlet:onboard)Before generating a challenge, register the in-loop variation provider so we do not call out to the Anthropic API just to spawn a sibling Claude (issue #464). Outside Claude Code this is a no-op and the default Anthropic provider remains active.
pythonfrom gauntlet.providers.in_loop import ( register_in_loop_provider_if_inside_claude_code, ) register_in_loop_provider_if_inside_claude_code()
.gauntlet/knowledge.json and developerprogress
.gauntlet/state/pending_challenge.json exists, evaluate the developer's most recent message as an answer before generating a new one
knowledge entry and challenge type
next challenge if in session.
new challenge.
| Result | Score | Streak | |--------|-------|--------| | Pass | 1.0 | +1 | | Partial | 0.5 | reset | | Fail | 0.0 | reset |
.gauntlet/knowledge.json exists and is readable before achallenge is generated; if missing, the skill surfaces the error and suggests running gauntlet:extract
to the developer's progress store and a streak update applied
.gauntlet/state/pending_challenge.json is evaluated beforegenerating a new challenge when it exists; the file is removed or updated after evaluation
the next challenge is presented (not skipped silently)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,309 | 4,510 | +5% | 1 | 1 | 0% | 633 | 729 | +15% | 0 | 0 | — |
case-02 | fail→fail | 8,990 | 9,593 | +7% | 1 | 1 | 0% | 1,440 | 992 | -31% | 0 | 0 | — |
case-03 | fail→fail | 12,338 | 6,155 | -50% | 1 | 1 | 0% | 776 | 798 | +3% | 0 | 0 | — |
case-04 | fail→pass | 11,629 | 2,182 | -81% | 1 | 1 | 0% | 1,780 | 920 | -48% | 0 | 0 | — |
case-05 | fail→pass | 11,891 | 1,681 | -86% | 1 | 1 | 0% | 1,843 | 811 | -56% | 0 | 0 | — |
case-06 | fail→pass | 15,298 | 2,984 | -80% | 1 | 1 | 0% | 2,464 | 1,006 | -59% | 0 | 0 | — |
case-07 | fail→pass | 11,603 | 2,945 | -75% | 1 | 1 | 0% | 1,771 | 1,124 | -37% | 0 | 0 | — |
case-08 | fail→pass | 9,779 | 2,165 | -78% | 1 | 1 | 0% | 1,495 | 951 | -36% | 0 | 0 | — |
case-09 | fail→pass | 7,504 | 1,343 | -82% | 1 | 1 | 0% | 1,295 | 757 | -42% | 0 | 0 | — |
case-10 | pass→pass | 11,466 | 1,775 | -85% | 1 | 1 | 0% | 1,774 | 817 | -54% | 0 | 0 | — |
case-11 | pass→pass | 9,189 | 2,002 | -78% | 1 | 1 | 0% | 1,465 | 817 | -44% | 0 | 0 | — |
case-12 | fail→fail | 9,902 | 1,794 | -82% | 1 | 1 | 0% | 1,540 | 791 | -49% | 0 | 0 | — |
case-13 | fail→pass | 5,426 | 2,165 | -60% | 1 | 1 | 0% | 816 | 859 | +5% | 0 | 0 | — |
case-14 | pass→pass | 8,763 | 2,222 | -75% | 1 | 1 | 0% | 1,287 | 833 | -35% | 0 | 0 | — |
case-15 | fail→fail | 12,038 | 2,551 | -79% | 1 | 1 | 0% | 1,698 | 879 | -48% | 0 | 0 | — |
case-16 | fail→pass | 12,798 | 7,522 | -41% | 1 | 1 | 0% | 1,992 | 1,589 | -20% | 0 | 0 | — |
case-17 | fail→pass | 9,842 | 1,943 | -80% | 1 | 1 | 0% | 1,429 | 776 | -46% | 0 | 0 | — |
case-18 | fail→pass | 10,747 | 2,387 | -78% | 1 | 1 | 0% | 1,726 | 935 | -46% | 0 | 0 | — |
case-19 | pass→pass | 11,704 | 4,131 | -65% | 1 | 1 | 0% | 1,927 | 1,232 | -36% | 0 | 0 | — |
case-20 | fail→pass | 11,705 | 2,759 | -76% | 1 | 1 | 0% | 1,781 | 935 | -48% | 0 | 0 | — |
case-21 | fail→fail | 11,477 | 2,335 | -80% | 1 | 1 | 0% | 1,801 | 883 | -51% | 0 | 0 | — |
case-22 | fail→fail | 10,184 | 2,321 | -77% | 1 | 1 | 0% | 1,729 | 946 | -45% | 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 19 counted toward the lift figure. The other 3 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 +50 percentage points is the difference between those two pass rates over the 19 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.