Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Reviews Godot TileMap implementation for known pitfalls. Triggers AFTER implementation, when code involves TileMapLayer, TileSet, TileSetAtlasSource, TileSetScenesCollectionSource, terrain painting, set_cell, erase_cell, get_cell_tile_data, collision polygons on tiles, or NavigationRegion2D with tile-based navigation. Do NOT use this skill for planning or teaching — only for post-implementation review.
.claude/skills/randallliuxin-tilemap-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -66% | 0% |
Post-implementation reviewer for Godot TileMap code. Checks against known gotchas that LLMs consistently get wrong.
After TileMap-related code is written or modified. Look for:
set_cell, erase_cell, set_cells_terrain_connect)get_cell_tile_data)y_sort_enabled on TileMapLayergotchas.mdchecklist.md static checks for automated verificationWhen you need specific API details, delegate to the godot-api skill.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→pass | 16,854 | 11,158 | -34% | 1 | 1 | 0% | 2,547 | 1,965 | -23% | 0 | 0 | — |
case-01 | fail→fail | 6,321 | 7,108 | +12% | 1 | 1 | 0% | 950 | 1,361 | +43% | 0 | 0 | — |
case-02 | fail→fail | 12,145 | 5,898 | -51% | 1 | 1 | 0% | 1,982 | 1,183 | -40% | 0 | 0 | — |
case-03 | fail→fail | 8,944 | 8,065 | -10% | 1 | 1 | 0% | 1,515 | 1,468 | -3% | 0 | 0 | — |
case-04 | fail→fail | 5,565 | 7,111 | +28% | 1 | 1 | 0% | 945 | 1,602 | +70% | 0 | 0 | — |
case-05 | fail→pass | 14,357 | 6,446 | -55% | 1 | 1 | 0% | 2,481 | 1,409 | -43% | 0 | 0 | — |
case-06 | pass→pass | 6,846 | 3,540 | -48% | 1 | 1 | 0% | 1,177 | 893 | -24% | 0 | 0 | — |
case-07 | fail→pass | 6,219 | 1,693 | -73% | 1 | 1 | 0% | 918 | 485 | -47% | 0 | 0 | — |
case-08 | fail→pass | 13,255 | 3,061 | -77% | 1 | 1 | 0% | 1,939 | 672 | -65% | 0 | 0 | — |
case-09 | fail→pass | 8,506 | 1,677 | -80% | 1 | 1 | 0% | 1,284 | 451 | -65% | 0 | 0 | — |
case-10 | pass→pass | 11,869 | 9,611 | -19% | 1 | 1 | 0% | 2,038 | 2,005 | -2% | 0 | 0 | — |
case-11 | pass→pass | 11,396 | 6,264 | -45% | 1 | 1 | 0% | 1,858 | 1,312 | -29% | 0 | 0 | — |
case-12 | pass→pass | 8,392 | 5,476 | -35% | 1 | 1 | 0% | 1,364 | 1,046 | -23% | 0 | 0 | — |
case-13 | pass→pass | 7,852 | 6,668 | -15% | 1 | 1 | 0% | 1,158 | 1,276 | +10% | 0 | 0 | — |
case-15 | pass→pass | 14,710 | 2,303 | -84% | 1 | 1 | 0% | 2,299 | 548 | -76% | 0 | 0 | — |
case-16 | pass→pass | 14,160 | 3,415 | -76% | 1 | 1 | 0% | 1,957 | 700 | -64% | 0 | 0 | — |
case-17 | pass→fail | 21,570 | 14,158 | -34% | 1 | 1 | 0% | 3,251 | 2,508 | -23% | 0 | 0 | — |
case-18 | pass→pass | 9,764 | 6,529 | -33% | 1 | 1 | 0% | 1,483 | 1,237 | -17% | 0 | 0 | — |
case-19 | fail→pass | 9,727 | 1,523 | -84% | 1 | 1 | 0% | 1,356 | 458 | -66% | 0 | 0 | — |
case-20 | pass→pass | 10,146 | 6,723 | -34% | 1 | 1 | 0% | 1,760 | 1,456 | -17% | 0 | 0 | — |
case-21 | pass→pass | 9,822 | 7,080 | -28% | 1 | 1 | 0% | 1,648 | 1,439 | -13% | 0 | 0 | — |
case-22 | pass→pass | 11,560 | 5,727 | -50% | 1 | 1 | 0% | 1,908 | 1,147 | -40% | 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. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.