Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a request asks to reference, refresh, write, or register project development context so the agent resolves the topic through agent_context/MAP.yaml and reads or updates the mapped Markdown context files.
.claude/skills/dexforce-project-dev-context/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -63% | 0% |
Canonical source: .agents/skills/project-dev-context/
reference, refresh, write, or register project development context.
.agents/skills/project-dev-context/SKILL.md.agent_context/MAP.yaml.Keep this file thin. If canonical routing behavior changes, update the canonical skill first, then only adjust this adapter if Claude needs a different local entry hint.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 7,811 | 5,462 | -30% | 1 | 1 | 0% | 922 | 538 | -42% | 0 | 0 | — |
case-02 | fail→fail | 18,360 | 6,228 | -66% | 1 | 1 | 0% | 3,042 | 689 | -77% | 0 | 0 | — |
case-22 | pass→pass | 8,992 | 7,011 | -22% | 1 | 1 | 0% | 1,639 | 1,593 | -3% | 0 | 0 | — |
case-03 | fail→fail | 8,592 | 6,088 | -29% | 1 | 1 | 0% | 1,437 | 529 | -63% | 0 | 0 | — |
case-04 | fail→pass | 13,014 | 6,664 | -49% | 1 | 1 | 0% | 2,197 | 1,465 | -33% | 0 | 0 | — |
case-05 | fail→fail | 14,188 | 6,129 | -57% | 1 | 1 | 0% | 2,403 | 1,278 | -47% | 0 | 0 | — |
case-06 | fail→pass | 16,420 | 15,437 | -6% | 1 | 1 | 0% | 2,949 | 2,974 | +1% | 0 | 0 | — |
case-07 | fail→pass | 12,548 | 6,667 | -47% | 1 | 1 | 0% | 1,979 | 1,417 | -28% | 0 | 0 | — |
case-08 | fail→pass | 9,491 | 2,991 | -68% | 1 | 1 | 0% | 1,589 | 726 | -54% | 0 | 0 | — |
case-09 | fail→fail | 11,017 | 5,249 | -52% | 1 | 1 | 0% | 1,990 | 524 | -74% | 0 | 0 | — |
case-10 | fail→pass | 11,491 | 2,763 | -76% | 1 | 1 | 0% | 1,819 | 666 | -63% | 0 | 0 | — |
case-11 | fail→fail | 14,884 | 7,240 | -51% | 1 | 1 | 0% | 2,387 | 1,361 | -43% | 0 | 0 | — |
case-12 | fail→fail | 14,719 | 5,468 | -63% | 1 | 1 | 0% | 2,738 | 708 | -74% | 0 | 0 | — |
case-13 | fail→fail | 10,774 | 2,842 | -74% | 1 | 1 | 0% | 1,926 | 587 | -70% | 0 | 0 | — |
case-14 | fail→pass | 16,559 | 6,148 | -63% | 1 | 1 | 0% | 2,843 | 1,017 | -64% | 0 | 0 | — |
case-15 | fail→pass | 11,735 | 2,950 | -75% | 1 | 1 | 0% | 1,939 | 810 | -58% | 0 | 0 | — |
case-16 | fail→fail | 14,592 | 7,947 | -46% | 1 | 1 | 0% | 2,368 | 621 | -74% | 0 | 0 | — |
case-17 | fail→pass | 10,493 | 2,552 | -76% | 1 | 1 | 0% | 1,609 | 666 | -59% | 0 | 0 | — |
case-18 | fail→fail | 13,532 | 17,560 | +30% | 1 | 1 | 0% | 2,422 | 582 | -76% | 0 | 0 | — |
case-19 | fail→fail | 102,373 | 5,462 | -95% | 1 | 1 | 0% | 2,132 | 523 | -75% | 0 | 0 | — |
case-20 | pass→pass | 11,945 | 8,799 | -26% | 1 | 1 | 0% | 2,441 | 2,011 | -18% | 0 | 0 | — |
case-21 | pass→fail | 5,281 | 4,639 | -12% | 1 | 1 | 0% | 972 | 1,021 | +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. 22 cases were attempted, and 16 counted toward the lift figure. The other 6 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 +32 percentage points is the difference between those two pass rates over the 16 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 | 8/22/2026 | +18% |
Other measured skills in the registry, with their headline benchmark lift.