Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use this skill when you learn one or more design pattern(s) in the Langroid (multi) agent framework, and want to make a note for future reference for yourself. Use this either autonomously, or when asked by the user to record a new pattern.
.claude/skills/pchalasani-add-pattern/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -15% | 0% |
When you learn a new Langroid design pattern, do the following:
patterns/SKILL.md file in the appropriate categorysection, containing a DESCRIPTION of the goal of the pattern (i.e. what it enables you to implement), accompanied by a - Reference: pointer to a markdown DOCUMENT in the patterns/ directory.
IMPORTANT - The DESCRIPTION should be clear enough that future YOU can effectively use it to MATCH design problems you may encounter in future.
Follow the format of existing pattern files (Problem, Solution, Complete Code Example, Key Points, When to Use).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→fail | 22,389 | 3,791 | -83% | 1 | 1 | 0% | 4,046 | 503 | -88% | 0 | 0 | — |
case-13 | fail→pass | 18,916 | 13,285 | -30% | 1 | 1 | 0% | 3,371 | 2,441 | -28% | 0 | 0 | — |
case-01 | fail→fail | 8,367 | 2,585 | -69% | 1 | 1 | 0% | 1,269 | 515 | -59% | 0 | 0 | — |
case-02 | fail→pass | 20,065 | 13,728 | -32% | 1 | 1 | 0% | 3,488 | 2,228 | -36% | 0 | 0 | — |
case-03 | fail→fail | 18,380 | 7,584 | -59% | 1 | 1 | 0% | 3,335 | 779 | -77% | 0 | 0 | — |
case-04 | fail→pass | 12,453 | 17,083 | +37% | 1 | 1 | 0% | 2,015 | 2,644 | +31% | 0 | 0 | — |
case-11 | pass→pass | 14,157 | 15,228 | +8% | 1 | 1 | 0% | 2,402 | 2,890 | +20% | 0 | 0 | — |
case-05 | fail→fail | 17,665 | 5,500 | -69% | 1 | 1 | 0% | 3,004 | 305 | -90% | 0 | 0 | — |
case-06 | fail→pass | 15,624 | 15,505 | -1% | 1 | 1 | 0% | 3,013 | 2,332 | -23% | 0 | 0 | — |
case-07 | fail→fail | 14,338 | 4,027 | -72% | 1 | 1 | 0% | 2,309 | 473 | -80% | 0 | 0 | — |
case-08 | fail→pass | 16,588 | 16,056 | -3% | 1 | 1 | 0% | 2,607 | 2,228 | -15% | 0 | 0 | — |
case-09 | fail→fail | 31,854 | 5,735 | -82% | 1 | 1 | 0% | 2,993 | 439 | -85% | 0 | 0 | — |
case-10 | fail→pass | 25,518 | 12,030 | -53% | 1 | 1 | 0% | 4,807 | 2,222 | -54% | 0 | 0 | — |
case-14 | fail→pass | 22,069 | 16,949 | -23% | 1 | 1 | 0% | 4,126 | 3,122 | -24% | 0 | 0 | — |
case-15 | fail→pass | 21,471 | 12,878 | -40% | 1 | 1 | 0% | 3,608 | 2,347 | -35% | 0 | 0 | — |
case-16 | fail→pass | 18,167 | 21,008 | +16% | 1 | 1 | 0% | 3,470 | 4,304 | +24% | 0 | 0 | — |
case-17 | fail→pass | 12,941 | 18,200 | +41% | 1 | 1 | 0% | 2,379 | 3,811 | +60% | 0 | 0 | — |
case-18 | pass→fail | 15,484 | 3,719 | -76% | 1 | 1 | 0% | 2,802 | 320 | -89% | 0 | 0 | — |
case-19 | fail→fail | 20,555 | 5,180 | -75% | 1 | 1 | 0% | 3,649 | 281 | -92% | 0 | 0 | — |
case-20 | pass→pass | 9,425 | 6,832 | -28% | 1 | 1 | 0% | 1,616 | 1,391 | -14% | 0 | 0 | — |
case-21 | fail→fail | 2,697 | 3,233 | +20% | 1 | 1 | 0% | 327 | 321 | -2% | 0 | 0 | — |
case-22 | pass→pass | 18,411 | 14,216 | -23% | 1 | 1 | 0% | 4,104 | 2,908 | -29% | 0 | 0 | — |
case-23 | pass→pass | 8,857 | 5,985 | -32% | 1 | 1 | 0% | 1,658 | 1,233 | -26% | 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 +35 percentage points is the difference between those two pass rates over the 18 comparable cases. 5 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.