Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Work on the core package (types, validation, normalization, diff). Use when modifying DSL processing logic, adding new node/edge properties, changing validation rules, or updating the diff algorithm.
.claude/skills/aiskillstore-core-development/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -10% | 0% |
The core package (packages/core/) is dependency-free and handles all DSL processing.
DSL (YAML input) → validate() → normalize() → IR → diff() → Patch| File | Purpose | Exports | |------|---------|---------| | types.ts | Type definitions | DSL, IR, Patch, WebSocket protocol | | validate.ts | YAML validation | validate(dsl): ValidationResult | | normalize.ts | DSL → IR conversion | normalize(dsl): IRDocument | | diff.ts | IR diff calculation | diff(prev, next): Patch |
DSL Types (user input) IR Types (normalized)
───────────────────── ────────────────────
DSLDocument IRDocument
├─ version: number ├─ version: number
├─ docId: string ├─ docId: string
├─ title?: string ├─ title: string
├─ nodes: DSLNode[] ├─ nodes: Record<string, IRNode>
└─ edges?: DSLEdge[] └─ edges: Record<string, IREdge>
DSLNode IRNode
├─ id: string ├─ id: string
├─ provider: string ├─ provider: string
├─ kind: string ├─ kind: string
├─ label?: string ├─ label: string (default: id)
├─ parent?: string ├─ parent: string | null
└─ layout: DSLLayout └─ layout: { x, y, w, h }
DSLEdge IREdge
├─ id: string ├─ id: string
├─ from: string ├─ from: string
├─ to: string ├─ to: string
└─ label?: string └─ label: string (default: "")typescripttype PatchOp = | { op: "upsertNode"; node: IRNode } | { op: "removeNode"; id: string } | { op: "upsertEdge"; edge: IREdge } | { op: "removeEdge"; id: string }; interface Patch { baseRev: number; nextRev: number; ops: PatchOp[]; }
typescript// Plugin → CLI interface HelloMessage { type: "hello"; docId: string; secret?: string; } interface RequestFullMessage { type: "requestFull"; docId: string; } // CLI → Plugin interface FullMessage { type: "full"; rev: number; ir: IRDocument; } interface PatchMessage { type: "patch"; baseRev: number; nextRev: number; ops: PatchOp[]; } interface ErrorMessage { type: "error"; message: string; }
types.tsvalidate.ts catches invalid inputnormalize.tsdiff.ts*.test.ts filesbun test packages/core/bash# All core tests bun test packages/core/ # Specific test file bun test packages/core/src/diff.test.ts bun test packages/core/src/validate.test.ts bun test packages/core/src/normalize.test.ts # Watch mode bun test --watch packages/core/
DSLNode and IRNode in types.tsvalidate.tsnormalize.tsDSLEdge and IREdge in types.tsvalidate.tsnormalize.ts| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 14,513 | 8,013 | -45% | 1 | 1 | 0% | 2,482 | 2,680 | +8% | 0 | 0 | — |
case-02 | fail→pass | 17,009 | 6,544 | -62% | 1 | 1 | 0% | 2,807 | 2,330 | -17% | 0 | 0 | — |
case-03 | fail→pass | 15,172 | 7,987 | -47% | 1 | 1 | 0% | 2,436 | 2,486 | +2% | 0 | 0 | — |
case-04 | fail→pass | 12,992 | 6,373 | -51% | 1 | 1 | 0% | 2,041 | 1,984 | -3% | 0 | 0 | — |
case-05 | pass→pass | 9,756 | 2,180 | -78% | 1 | 1 | 0% | 1,725 | 1,365 | -21% | 0 | 0 | — |
case-06 | fail→pass | 11,616 | 4,405 | -62% | 1 | 1 | 0% | 1,915 | 1,733 | -10% | 0 | 0 | — |
case-07 | fail→fail | 12,434 | 4,689 | -62% | 1 | 1 | 0% | 1,952 | 1,813 | -7% | 0 | 0 | — |
case-08 | fail→pass | 14,074 | 4,459 | -68% | 1 | 1 | 0% | 2,231 | 1,721 | -23% | 0 | 0 | — |
case-09 | pass→pass | 9,262 | 1,508 | -84% | 1 | 1 | 0% | 1,434 | 1,198 | -16% | 0 | 0 | — |
case-10 | fail→pass | 12,864 | 2,012 | -84% | 1 | 1 | 0% | 2,030 | 1,313 | -35% | 0 | 0 | — |
case-11 | pass→pass | 11,526 | 2,094 | -82% | 1 | 1 | 0% | 1,789 | 1,297 | -28% | 0 | 0 | — |
case-12 | pass→pass | 9,137 | 2,149 | -76% | 1 | 1 | 0% | 1,464 | 1,345 | -8% | 0 | 0 | — |
case-13 | pass→pass | 13,448 | 3,549 | -74% | 1 | 1 | 0% | 2,115 | 1,590 | -25% | 0 | 0 | — |
case-14 | fail→pass | 9,023 | 2,899 | -68% | 1 | 1 | 0% | 1,424 | 1,432 | +1% | 0 | 0 | — |
case-15 | fail→pass | 15,061 | 1,591 | -89% | 1 | 1 | 0% | 2,501 | 1,284 | -49% | 0 | 0 | — |
case-16 | fail→pass | 16,163 | 1,968 | -88% | 1 | 1 | 0% | 2,585 | 1,275 | -51% | 0 | 0 | — |
case-17 | fail→pass | 21,840 | 2,740 | -87% | 1 | 1 | 0% | 2,547 | 1,442 | -43% | 0 | 0 | — |
case-18 | fail→pass | 17,093 | 2,366 | -86% | 1 | 1 | 0% | 2,177 | 1,364 | -37% | 0 | 0 | — |
case-19 | fail→pass | 15,051 | 3,154 | -79% | 1 | 1 | 0% | 2,296 | 1,538 | -33% | 0 | 0 | — |
case-20 | fail→pass | 7,914 | 1,712 | -78% | 1 | 1 | 0% | 1,248 | 1,252 | +0% | 0 | 0 | — |
case-21 | pass→pass | 11,294 | 2,475 | -78% | 1 | 1 | 0% | 1,926 | 1,405 | -27% | 0 | 0 | — |
case-22 | pass→pass | 16,946 | 2,052 | -88% | 1 | 1 | 0% | 2,733 | 1,284 | -53% | 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 +64 percentage points is the difference between those two pass rates over the 22 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.