Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/w1ne/kernelcad-web/kernelcad-assembliesnpx skills add w1ne/kernelCAD-web --skill kernelcad-assembliesgit clone --depth 1 https://github.com/w1ne/kernelCAD-webWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00050 | $0.14352 |
| Opus 5 | $0.00025 | $0.07176 |
| Sonnet 5 | $0.00010 | $0.02870 |
| Haiku 4.5 | $0.00005 | $0.01435 |
Grade A, and why
kernelcad-assemblies scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
arm.mate('left-curl', 'palm.left-hinge', 'left-finger.hinge', 'revolute'); How it starts
The opening of the file, as written. The whole thing — 778 lines — stays where its author put it; the contents beside it link to each section on GitHub.
kernelCAD — assemblies
Use assembly() when the model needs named mechanical parts, connector frames, and joint metadata that a human or agent can inspect later. Call .model() after adding parts to return a Scene with per-part bodies; iterate .parts for per-part rendering / export, call .toCompound() for an OCCT group (lossless on color/name/identity, default for STEP), or .toUnion() for one fused solid (lossy on color/name — antipattern except when downstream truly needs a single Shape). Connector and joint records remain metadata for now; .model() does not solve motion (use .solvedModel(poses) for that). Use MCP inspect({ of: 'assemblies', file? code? }) to inspect the captured assembly intent without recomputing topology.
Assembly validity
kernelcad validate <file.kcad.ts> runs the assembly validator over the script's Scene. Three checks today:
assembly.part.floating— a part has no joint connecting it to any other part. The fix: declare the connection viaarm.mate(..., 'fastened')orarm.mate(..., 'revolute', ...).assembly.part.orphan— a part is in a sub-assembly disconnected from the main mechanism.assembly.interference.overlap— two parts share volume (promoted fromkernelcad interference).assembly.structure.unstructured-bodies(info) — a multi-body model returns loose top-level bodies with noassembly().part(...)structure, so the parts carry no identity forinspect --focus,inspect({ of: 'part-stats' }), or per-part review. The fix: wrap each distinct body in a namedassembly().part(name, shape).
Exit codes: 0 (solved) / 1 (warnings) / 2 (errors). Pipe-friendly:
kernelcad validate so100.kcad.ts && echo "fits"
Authoring rule: every arm.part(name, shape) should also appear in at least one arm.mate(name, 'a.connector', 'b.connector', kind, opts) call (kind ∈ fastened | revolute | prismatic | ball). Raw at: [x, y, z] placement without a mate produces a working geometric output but fails validation — the agent has authored a position but not a connection.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 778 lines · 50 tokens per session scan A 362f8edea0ff
kernelcad-assemblies is a skill published in the GitHub repository w1ne/kernelCAD-web (17 stars, last pushed 4d ago), licensed MIT. It adds 50 tokens to every session and 14,352 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
simplecadapi
Thin SimpleCAD SDK reference skill focused on the public API surface, core types, and current modeling workflows.
opengis-skills
Use when AI coding assistant needs GIS/CAD/C#/AI/IoT/3D domain expertise for 72+ open-source projects. One-stop skill index with tag-based search and on-demand loading for GDAL, GeoServer, QGIS, PostGIS, JTS, CesiumJS, FreeCAD, OpenSCAD, OCCT, NPOI, SqlSugar, Furion, Dify, SuperSplat, Go and more.
cad
Use when doing parametric 3D modeling, 2D drafting, geometric kernel development, BIM/IFC processing, PCB design, or AutoCAD .NET development. Index of 19 skills: FreeCAD, OpenSCAD, OCCT, CadQuery, KiCad, SolveSpace, QCAD, xBIM, Clipper2, TongWen and more.
cadquery-modeling
Generate CAD models with CadQuery using an adaptive CAD Compiler workflow: plan-mode skill constraint handoff, focused iteration scopes, fail-first challenge review, compact review packets, real-product reference fidelity, visual/functional audits, source-first repair, and STEP/STL export. Use for CadQuery, CQ, 建模…
aieng-cad-authoring
Active MCP-first CAD authoring skill for AIENG Workbench. Use when the user asks to create, design, draft, model, or iteratively build CAD geometry through the active cad. MCP tools. Do not use for solver/result review, schema/library implementation, or editing legacy IR plans.
aieng-cad-cae-copilot
MCP-first evidence workflow for AIENG CAD/CAE setup, solver preflight, approval-gated solver execution, and result extraction. Use when the user asks to inspect CAE readiness, patch setup, prepare/run CalculiX, or report stress/displacement evidence.