Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan and execute safe refactoring using dependency analysis
.claude/skills/tirth8205-refactor-safely/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-23 | ✗→✓ | ▲ Improved | -64% | 0% |
Plan a refactor from the dependency graph and apply renames from a preview.
refactor_tool with mode="suggest" for refactoring candidates, or mode="dead_code" for unreferenced code.refactor_tool with mode="rename", old_name and new_name. Check the returned edit list before applying.apply_refactor_tool with the returned refactor_id to apply the rename.get_impact_radius_tool and get_affected_flows_tool to see the dependents and critical paths involved.find_large_functions_tool to find functions worth splitting.detect_changes_tool to confirm the impact matches the plan.get_minimal_context_tool(task="<your task>") before any other graph tool.detail_level="minimal" wherever a tool accepts it. Use "standard" only when minimal is not enough.query_graph_tool call over a broad listing call.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | pass→pass | 16,308 | 2,024 | -88% | 1 | 1 | 0% | 1,695 | 543 | -68% | 0 | 0 | — |
case-01 | fail→fail | 4,017 | 17,828 | +344% | 1 | 1 | 0% | 434 | 786 | +81% | 0 | 0 | — |
case-16 | pass→pass | 18,965 | 8,674 | -54% | 1 | 1 | 0% | 1,576 | 766 | -51% | 0 | 0 | — |
case-02 | fail→fail | 8,321 | 6,137 | -26% | 1 | 1 | 0% | 190 | 605 | +218% | 0 | 0 | — |
case-03 | fail→fail | 19,365 | 11,544 | -40% | 1 | 1 | 0% | 2,055 | 590 | -71% | 0 | 0 | — |
case-04 | fail→fail | 14,354 | 7,274 | -49% | 1 | 1 | 0% | 1,221 | 569 | -53% | 0 | 0 | — |
case-05 | fail→pass | 18,142 | 2,556 | -86% | 1 | 1 | 0% | 2,031 | 636 | -69% | 0 | 0 | — |
case-06 | pass→pass | 7,978 | 7,494 | -6% | 1 | 1 | 0% | 1,076 | 601 | -44% | 0 | 0 | — |
case-07 | fail→fail | 9,323 | 3,808 | -59% | 1 | 1 | 0% | 536 | 791 | +48% | 0 | 0 | — |
case-08 | fail→fail | 16,897 | 7,826 | -54% | 1 | 1 | 0% | 1,271 | 687 | -46% | 0 | 0 | — |
case-09 | fail→fail | 10,596 | 7,718 | -27% | 1 | 1 | 0% | 1,645 | 684 | -58% | 0 | 0 | — |
case-10 | fail→pass | 13,153 | 11,835 | -10% | 1 | 1 | 0% | 1,176 | 782 | -34% | 0 | 0 | — |
case-11 | fail→pass | 16,148 | 10,830 | -33% | 1 | 1 | 0% | 2,629 | 1,036 | -61% | 0 | 0 | — |
case-12 | fail→fail | 6,645 | 7,605 | +14% | 1 | 1 | 0% | 900 | 611 | -32% | 0 | 0 | — |
case-13 | fail→fail | 16,098 | 3,752 | -77% | 1 | 1 | 0% | 1,427 | 826 | -42% | 0 | 0 | — |
case-14 | fail→fail | 14,703 | 9,606 | -35% | 1 | 1 | 0% | 1,460 | 813 | -44% | 0 | 0 | — |
case-17 | pass→pass | 17,020 | 2,420 | -86% | 1 | 1 | 0% | 1,738 | 508 | -71% | 0 | 0 | — |
case-18 | pass→pass | 14,399 | 9,872 | -31% | 1 | 1 | 0% | 1,909 | 748 | -61% | 0 | 0 | — |
case-19 | fail→pass | 13,861 | 7,838 | -43% | 1 | 1 | 0% | 1,162 | 697 | -40% | 0 | 0 | — |
case-20 | pass→fail | 20,282 | 16,580 | -18% | 1 | 1 | 0% | 2,322 | 552 | -76% | 0 | 0 | — |
case-21 | pass→fail | 6,130 | 16,556 | +170% | 1 | 1 | 0% | 717 | 533 | -26% | 0 | 0 | — |
case-22 | pass→fail | 20,187 | 16,616 | -18% | 1 | 1 | 0% | 2,880 | 564 | -80% | 0 | 0 | — |
case-23 | fail→pass | 16,666 | 7,364 | -56% | 1 | 1 | 0% | 1,480 | 528 | -64% | 0 | 0 | — |
case-24 | fail→fail | 20,284 | 11,869 | -41% | 1 | 1 | 0% | 1,826 | 742 | -59% | 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. 24 cases were attempted, and 18 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 +8 percentage points is the difference between those two pass rates over the 18 comparable cases. 3 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/24/2026 | +36% |
| gemini-3.6-flash | verified | 8/9/2026 | +48% |
Other measured skills in the registry, with their headline benchmark lift.