EDA Skillpack

All skills for all EDA capabilities across the Adom ecosystem.

This is an ecosystem-wide entry point for EDA skills, combining an index of skills maintained in other packages with shared guidance hosted here. Its scope includes component discovery, symbols, footprints, libraries, schematic and PCB design, routing, simulation and analysis, manufacturing outputs, fabrication, placement, soldering, reflow and inspection—across desktop bridges and Adom web tools, for any AI. The source and installable package are available. Install with adom-wiki pkg install adom/eda-engineering, then read eda-engineering and the relevant focused skill.

The initial skills cover current/load reasoning, routing geometry, thermal pads and vias, copper-ablation retention, thermal-bottleneck rerouting, and assembly planning. It contains AI-neutral Markdown and no CAD application dependency. These are engineering workflows, not automatic board certification or prequalified machine recipes.

The Astra/Fusion video progression supplied worked examples. Device assumptions and measured outcomes stay with each example; reusable decisions live here.

Browse the skill directory to find the relevant package or shared skill. The existing EDA hub remains the application directory.

Fusion, KiCad and Altium may keep their own skillpacks inside their bridges. Broad guidance, such as how to plan copper pours, can live here while each bridge explains how to execute it. A future Adom Molecule Skillpack could also be cross-linked; its packaging remains undecided. “All skills” describes the ecosystem-wide scope, not a requirement to move everything into one tarball. Links do not automatically install the packages they point to.

The showcase pairs Fusion, Altium and KiCad with Adom’s browser tools: adom-symbol for schematic symbols, adom-footprint for pads and paste geometry, adom-lbr for symbol/footprint/3D libraries and format conversion, and adom-parts-search for component sourcing. These are real application captures; the collage illustrates the broader workflow, not a claim that every tool has completed the Fusion routing demonstration.

Worked example: independent AI re-route vs the hand-routed original

Fable 5.1 vs Adonis on the BQ25792 charger: the same board routed from a proven-stripped copy by an AI's own planner (no autorouter), validated natively in Fusion, and compared net by net with the original in a two-minute side-by-side video. The page keeps the numbers, the boards, every DRC result and the planner's iteration history so other AIs can use it as reference data. Four shared skills came out of it: independent re-route comparison, native connectivity conformance, fine-pitch fan-out and pour planning and measurement.

Route visibly with shared engineering guidance

Ask your AI: “Use eda-visible-routing from the EDA Skillpack to route my working board copy visibly, with clean bends, electrical priorities, expanded copper pours and thermal bottleneck review. Use my selected EDA bridge for execution and native validation.”

Read the skill directly from the wiki. Install with adom-wiki pkg install adom/eda-engineering. It supports any AI and delegates tool-specific operations to the appropriate bridge.

Independent placement, routing, 3D review and video

Watch the narrated Astra/Fusion comparison: all components moved off-board, a fresh placement and routes, native verification, and a side-by-side 3D assembly comparison. The shared methods are AI-neutral and use each selected bridge for native execution.

• eda-placement-routing-review: Coordinate an independent PCB placement, trace routing, native 3D review and narrated comparison.
• eda-component-placement: Calculate PCB component placement from nets, footprints, power paths and mechanical interfaces.
• eda-board-3d-comparison: Compare original and redesigned PCB assemblies in the native EDA 3D viewer.
• eda-design-summary-video: Make a narrated evidence-based video of PCB placement, routing, validation and native 3D comparison.

These skills are published as readable source. The installable package is still pending; read the files directly through the wiki rather than assume an installation command succeeds.

KiCad worked example: multilayer power and copper retention

The Codex Astra + KiCad guide follows simple live traces through a public four-layer BMS. The later BMS X-ray and current review compares matched nets, revisits power vias and supply-plane stitching, and separates floating copper retention from useful electrical copper.

The general lessons are in eda-multilayer-current-review, linked from the copper-ablation and pour-measurement skills. They apply to any AI and EDA tool: explicit load terminals, native refilled geometry, conductor connectivity, numerical verification, honest legends and validation of the actual saved revision. These are analysis graphics, not a claim that every bridge has a native X-ray or current-field verb. Device-specific values and unresolved checks remain with the worked example.

Read these skills directly from the source or install the published EDA Skillpack. Adding source guidance does not update already-installed packages or instructions loaded in an existing AI conversation.

TI-based layout skills for Claude and Codex

• Motor-driver layout, based on TI SLVA959B.
• Kelvin current-sense layout, based on TI SSZTAD8.
• Addressable RGB LEDs, based on the Worldsemi 2020 datasheets and the LED Grid Wall build.

Both are AI-neutral Markdown, with source links, board-specific decision checks and native-refill verification. Read the same files from Claude or Codex. Their numerical examples are not universal board rules. The install script deploys identical files to both agent skill homes.

Independent Astra ESC source and measurements

Independent placement, routing and copper pours includes Python and C++ source, the clean fixture, accepted plans and native evidence. The algorithms illustrate reusable methods; the board-specific policy must be reviewed for each new design. These resources are available to any AI.