Parametric cabinet design toolkit with MCP server — cutlist, evaluation, and 3D visualization
Skill Claude CodeCodex
Add a new named cabinet preset to cabineteer. Use when asked to add a preset, template, or canned design to presets.py (surfaced via listpresets / applypreset). Covers the CabinetPreset registration pattern, the opening-stack sum invariant, and the guard test that every preset must evaluate with zero errors.
Skill Claude CodeCodex
Add a new eval scenario to cabineteer's eval suite. Use when covering a new tool behavior, a bug regression, or a workflow chain in evals/scenarios.py. Covers the Scenario/ToolCall/Assertion structure, the assertion operators, context chaining between calls, and the meta-tests every scenario must satisfy.
Skill Claude CodeCodex
Run Charlie's full-codebase or multi-module review-and-fix workflow for cabineteer. Use for a broad correctness/security review across many modules, or when asked for a "top-to-bottom" review. Covers the parallel module-agent fan-out, verification, the fixes-as-their-own-PR convention, and merging only on explicit go.
Skill Claude CodeCodex
Call cabineteer tool handlers directly from Python without the MCP transport or a live server. Use to exercise or debug a tool (designcabinet, evaluatecabinet, generatecutlist, visualizecabinet, autofixcabinet, etc.), reproduce an eval scenario, or check behavior when the session's cabineteer MCP server is stale…
Skill Claude CodeCodex
Run the cabineteer eval suite and interpret results. Use whenever you finish a non-trivial change, want to check the eval baseline (current counts live in CLAUDE.md and this skill's body), filter evals by tag/difficulty/name, or debug a failing scenario. The eval harness calls tool handlers directly (no MCP transport)…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: