Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Produce an implementation-ready plan for a software change by inspecting the actual repository, tracing affected execution paths, resolving material design decisions, and defining verification and rollout evidence. Use when the user asks for an implementation plan, technical approach, task breakdown, design proposal grounded in an existing codebase, or wants to know how a feature or fix should be built before editing begins. Do not use for generic product roadmaps or when the user has asked to i
.claude/skills/contextosai-plan-change/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 129% | 0% |
| case-05 | ✓→✗ | ▼ Worse | -66% | 0% |
| case-14 | ✓→✗ | ▼ Worse | -65% | 0% |
Create a plan another engineer can execute without rediscovering the system.
user changes.
Infer only low-risk details; surface choices that materially change scope, compatibility, data, security, or user experience.
wiring, domain logic, state, tests, configuration, and operational controls.
error, retry, authorization, concurrency, and rollback paths only when touched.
they lose in this repository; omit decorative option lists.
symbols, the logic change, its contract with adjacent code, and verification.
rollback steps when relevant. Never hide them under “update config.”
implementation and verification, and every risky mutation has a recovery path.
path:line evidence for the current architecture and chosen edit points.which behavior changes, where, and how it is proven.
arbitrary file-by-file edits.
Use references/implementation-plan.md. Lead with the approach and key decisions, then provide ordered implementation slices, verification, and rollout. If one unresolved decision would produce substantially different plans, stop and ask for it rather than pretending the branches are equivalent.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→fail | 11,970 | 3,674 | -69% | 1 | 1 | 0% | 2,024 | 588 | -71% | 0 | 0 | — |
case-01 | fail→fail | 26,817 | 2,976 | -89% | 1 | 1 | 0% | 6,083 | 591 | -90% | 0 | 0 | — |
case-02 | fail→fail | 20,157 | 2,266 | -89% | 1 | 1 | 0% | 4,322 | 604 | -86% | 0 | 0 | — |
case-03 | fail→fail | 4,136 | 4,533 | +10% | 1 | 1 | 0% | 154 | 613 | +298% | 0 | 0 | — |
case-04 | pass→fail | 7,594 | 18,354 | +142% | 1 | 1 | 0% | 1,822 | 4,167 | +129% | 0 | 0 | — |
case-05 | pass→fail | 10,150 | 3,015 | -70% | 1 | 1 | 0% | 2,221 | 756 | -66% | 0 | 0 | — |
case-06 | fail→fail | 6,274 | 3,428 | -45% | 1 | 1 | 0% | 1,302 | 598 | -54% | 0 | 0 | — |
case-07 | fail→fail | 14,079 | 3,255 | -77% | 1 | 1 | 0% | 3,679 | 581 | -84% | 0 | 0 | — |
case-08 | fail→fail | 6,454 | 3,443 | -47% | 1 | 1 | 0% | 1,315 | 786 | -40% | 0 | 0 | — |
case-09 | fail→pass | 8,580 | 12,895 | +50% | 1 | 1 | 0% | 1,789 | 3,110 | +74% | 0 | 0 | — |
case-10 | fail→fail | 24,727 | 3,955 | -84% | 1 | 1 | 0% | 6,179 | 626 | -90% | 0 | 0 | — |
case-11 | fail→fail | 14,254 | 3,255 | -77% | 1 | 1 | 0% | 3,193 | 610 | -81% | 0 | 0 | — |
case-12 | fail→fail | 14,127 | 3,603 | -74% | 1 | 1 | 0% | 2,742 | 670 | -76% | 0 | 0 | — |
case-13 | fail→fail | 7,208 | 19,491 | +170% | 1 | 1 | 0% | 1,740 | 4,311 | +148% | 0 | 0 | — |
case-14 | pass→fail | 9,747 | 4,004 | -59% | 1 | 1 | 0% | 1,937 | 670 | -65% | 0 | 0 | — |
case-15 | pass→pass | 10,552 | 22,432 | +113% | 1 | 1 | 0% | 1,937 | 4,132 | +113% | 0 | 0 | — |
case-16 | fail→pass | 20,179 | 23,993 | +19% | 1 | 1 | 0% | 3,902 | 4,674 | +20% | 0 | 0 | — |
case-18 | fail→fail | 15,294 | 2,233 | -85% | 1 | 1 | 0% | 3,452 | 596 | -83% | 0 | 0 | — |
case-19 | fail→fail | 11,598 | 3,718 | -68% | 1 | 1 | 0% | 2,491 | 618 | -75% | 0 | 0 | — |
case-20 | pass→fail | 9,777 | 3,623 | -63% | 1 | 1 | 0% | 1,601 | 556 | -65% | 0 | 0 | — |
case-21 | fail→fail | 13,206 | 4,419 | -67% | 1 | 1 | 0% | 2,280 | 647 | -72% | 0 | 0 | — |
case-22 | fail→fail | 9,907 | 3,676 | -63% | 1 | 1 | 0% | 1,979 | 605 | -69% | 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 5 counted toward the lift figure. The other 17 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 -9 percentage points is the difference between those two pass rates over the 5 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.