Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →KiCad EDA orchestration via kicad-mcp MCP server. Routes 17 tools for schematic creation, PCB layout, autorouting, DRC, and Gerber export. Enforces serialized PCB ops, library-first lookup, and autoroute-only routing.
.claude/skills/telagod-operating-kicad-eda/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -26% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 1% | 0% |
> 判断先于执行：决定「是否做 / 选什么 / 如何取舍」（栈、方案、架构、权衡）前，先读领域判断内核 skills/_kernel/hardware/SKILL.md——它管 judgment，本秘典管 execution；冲突时以内核判断为准。
> MCP tool router for KiCad 9. Never guess, never hand-route, never parallelize PCB writes.
| Scenario | Use | Approach | |----------|-----|----------| | Create schematic from design doc | Yes | schematic() operations | | Generate PCB from schematic | Yes | build_pcb_from_schematic | | Route a PCB | Yes | autoroute(operation="run") only | | Fix DRC violations | Yes | drc(operation="autofix") | | Export manufacturing files | Yes | export() Gerber/BOM/render | | Hand-place a single trace | NO | LLM traces violate clearances |
pcb(add_trace/add_via) produces DRC-failing garbage. Use autoroute(operation="run").library(operation="search") first..kicad_pcb file.drc(operation="run") + audit(operation="all") before declaring done.| Task | MCP Tool | Operation | |------|----------|-----------| | Search parts | library | search | | Create/edit schematic | schematic | create_symbol, add_wire, add_label | | Build PCB from schematic | build_pcb_from_schematic | — (standalone) | | Place footprints | pcb | place_footprint | | Set board outline | pcb | set_board_outline | | Add zones (copper pour) | pcb | add_zone | | Auto-route | autoroute | run | | Run DRC | drc | run | | Auto-fix DRC | drc | autofix | | Export Gerber | export | gerber | | Export BOM | export | bom | | 3D render | export | render | | Estimate board size | estimate_board_size | — (standalone) | | Suggest placement | suggest_placement | — (standalone) | | Full audit | audit | all |
1. library(search) → find symbols and footprints
2. schematic(create_symbol) → place all components
3. schematic(add_wire/add_label) → connect nets
4. [Python] footprint injection script → fix kicad-cli lookup
5. build_pcb_from_schematic → initial PCB
6. pcb(set_board_outline) → product form factor
7. suggest_placement or pcb(place_footprint) → position components
8. autoroute(run) → FreeRouter
9. drc(run) → check
10. drc(autofix) → fix violations
11. pcb(add_zone) → GND copper pour both layers
12. export(gerber) → manufacturing files
13. export(render) → 3D views| Topic | File | |-------|------| | KiCad 9 parse quirks | kicad9-quirks.md | | DRC fix strategies | drc-strategies.md | | Footprint injection | footprint-injection.md |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 32,166 | 35,666 | +11% | 1 | 1 | 0% | 6,208 | 7,196 | +16% | 0 | 0 | — |
case-02 | fail→pass | 16,065 | 18,760 | +17% | 1 | 1 | 0% | 2,582 | 3,562 | +38% | 0 | 0 | — |
case-03 | fail→pass | 35,307 | 26,061 | -26% | 1 | 1 | 0% | 6,200 | 4,568 | -26% | 0 | 0 | — |
case-04 | pass→pass | 7,531 | 2,855 | -62% | 1 | 1 | 0% | 1,159 | 1,450 | +25% | 0 | 0 | — |
case-05 | fail→pass | 14,260 | 2,992 | -79% | 1 | 1 | 0% | 1,762 | 1,447 | -18% | 0 | 0 | — |
case-06 | pass→pass | 13,902 | 3,310 | -76% | 1 | 1 | 0% | 1,910 | 1,424 | -25% | 0 | 0 | — |
case-07 | pass→pass | 13,022 | 5,569 | -57% | 1 | 1 | 0% | 1,829 | 1,913 | +5% | 0 | 0 | — |
case-08 | pass→pass | 10,568 | 7,425 | -30% | 1 | 1 | 0% | 1,700 | 2,161 | +27% | 0 | 0 | — |
case-09 | fail→fail | 16,903 | 3,870 | -77% | 1 | 1 | 0% | 2,807 | 1,534 | -45% | 0 | 0 | — |
case-10 | fail→fail | 10,770 | 2,819 | -74% | 1 | 1 | 0% | 1,348 | 1,329 | -1% | 0 | 0 | — |
case-11 | fail→fail | 11,160 | 3,428 | -69% | 1 | 1 | 0% | 1,808 | 1,437 | -21% | 0 | 0 | — |
case-12 | fail→pass | 7,558 | 2,611 | -65% | 1 | 1 | 0% | 1,205 | 1,270 | +5% | 0 | 0 | — |
case-13 | pass→pass | 9,987 | 2,739 | -73% | 1 | 1 | 0% | 1,457 | 1,421 | -2% | 0 | 0 | — |
case-14 | fail→pass | 8,321 | 2,531 | -70% | 1 | 1 | 0% | 1,320 | 1,333 | +1% | 0 | 0 | — |
case-15 | fail→pass | 7,854 | 3,196 | -59% | 1 | 1 | 0% | 1,168 | 1,385 | +19% | 0 | 0 | — |
case-16 | pass→pass | 10,501 | 5,358 | -49% | 1 | 1 | 0% | 1,679 | 1,661 | -1% | 0 | 0 | — |
case-17 | fail→pass | 8,471 | 2,766 | -67% | 1 | 1 | 0% | 1,296 | 1,421 | +10% | 0 | 0 | — |
case-18 | fail→pass | 10,368 | 5,114 | -51% | 1 | 1 | 0% | 1,576 | 1,636 | +4% | 0 | 0 | — |
case-19 | pass→pass | 12,717 | 2,039 | -84% | 1 | 1 | 0% | 1,860 | 1,268 | -32% | 0 | 0 | — |
case-20 | pass→pass | 12,759 | 3,185 | -75% | 1 | 1 | 0% | 1,885 | 1,493 | -21% | 0 | 0 | — |
case-21 | fail→pass | 10,536 | 6,496 | -38% | 1 | 1 | 0% | 1,543 | 1,838 | +19% | 0 | 0 | — |
case-22 | fail→pass | 57,016 | 16,583 | -71% | 1 | 1 | 0% | 1,292 | 3,759 | +191% | 0 | 0 | — |
case-23 | pass→pass | 3,879 | 4,706 | +21% | 1 | 1 | 0% | 753 | 1,836 | +144% | 0 | 0 | — |
case-24 | pass→pass | 10,619 | 7,257 | -32% | 1 | 1 | 0% | 1,928 | 2,395 | +24% | 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 23 counted toward the lift figure. The other 1 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 +42 percentage points is the difference between those two pass rates over the 23 comparable cases.
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.