Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate dynamic plugin loading system with discovery, validation, and lifecycle management.
.claude/skills/a5c-ai-plugin-loader-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -49% | 0% |
Generate dynamic plugin loading system.
typescriptimport fs from 'fs'; import path from 'path'; interface Plugin { name: string; version: string; init: () => Promise<void>; destroy?: () => Promise<void>; } export class PluginLoader { private plugins = new Map<string, Plugin>(); private pluginDirs: string[]; constructor(pluginDirs: string[]) { this.pluginDirs = pluginDirs; } async discover(): Promise<string[]> { const found: string[] = []; for (const dir of this.pluginDirs) { if (!fs.existsSync(dir)) continue; const entries = fs.readdirSync(dir, { withFileTypes: true }); for (const entry of entries) { if (entry.isDirectory()) { const manifestPath = path.join(dir, entry.name, 'manifest.json'); if (fs.existsSync(manifestPath)) { found.push(path.join(dir, entry.name)); } } } } return found; } async load(pluginPath: string): Promise<Plugin> { const manifest = JSON.parse(fs.readFileSync(path.join(pluginPath, 'manifest.json'), 'utf-8')); const module = await import(path.join(pluginPath, manifest.main)); const plugin: Plugin = { name: manifest.name, version: manifest.version, ...module }; await plugin.init(); this.plugins.set(plugin.name, plugin); return plugin; } async unload(name: string): Promise<void> { const plugin = this.plugins.get(name); if (plugin?.destroy) await plugin.destroy(); this.plugins.delete(name); } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 19,586 | 17,575 | -10% | 1 | 1 | 0% | 3,963 | 3,502 | -12% | 0 | 0 | — |
case-02 | fail→fail | 22,872 | 11,281 | -51% | 1 | 1 | 0% | 3,905 | 2,964 | -24% | 0 | 0 | — |
case-03 | fail→fail | 16,833 | 13,585 | -19% | 1 | 1 | 0% | 3,735 | 3,381 | -9% | 0 | 0 | — |
case-04 | fail→fail | 8,601 | 5,750 | -33% | 1 | 1 | 0% | 1,645 | 1,487 | -10% | 0 | 0 | — |
case-05 | fail→pass | 11,996 | 9,681 | -19% | 1 | 1 | 0% | 2,051 | 2,196 | +7% | 0 | 0 | — |
case-06 | pass→pass | 11,090 | 3,366 | -70% | 1 | 1 | 0% | 1,769 | 1,022 | -42% | 0 | 0 | — |
case-07 | pass→pass | 13,144 | 11,938 | -9% | 1 | 1 | 0% | 2,276 | 2,974 | +31% | 0 | 0 | — |
case-08 | fail→fail | 9,863 | 6,889 | -30% | 1 | 1 | 0% | 1,958 | 1,845 | -6% | 0 | 0 | — |
case-09 | fail→fail | 12,676 | 8,786 | -31% | 1 | 1 | 0% | 2,388 | 2,183 | -9% | 0 | 0 | — |
case-10 | fail→fail | 12,921 | 5,189 | -60% | 1 | 1 | 0% | 2,328 | 1,477 | -37% | 0 | 0 | — |
case-11 | fail→pass | 13,465 | 8,258 | -39% | 1 | 1 | 0% | 2,419 | 1,717 | -29% | 0 | 0 | — |
case-12 | fail→fail | 11,453 | 19,379 | +69% | 1 | 1 | 0% | 2,113 | 1,283 | -39% | 0 | 0 | — |
case-13 | fail→pass | 10,937 | 2,306 | -79% | 1 | 1 | 0% | 1,954 | 892 | -54% | 0 | 0 | — |
case-14 | fail→pass | 11,231 | 8,179 | -27% | 1 | 1 | 0% | 2,330 | 1,794 | -23% | 0 | 0 | — |
case-15 | fail→pass | 14,939 | 4,796 | -68% | 1 | 1 | 0% | 2,252 | 1,158 | -49% | 0 | 0 | — |
case-16 | fail→fail | 18,064 | 5,058 | -72% | 1 | 1 | 0% | 2,887 | 1,526 | -47% | 0 | 0 | — |
case-17 | fail→fail | 10,219 | 17,250 | +69% | 1 | 1 | 0% | 2,197 | 1,074 | -51% | 0 | 0 | — |
case-18 | fail→pass | 12,648 | 4,692 | -63% | 1 | 1 | 0% | 2,446 | 1,373 | -44% | 0 | 0 | — |
case-19 | fail→pass | 9,541 | 2,564 | -73% | 1 | 1 | 0% | 1,513 | 973 | -36% | 0 | 0 | — |
case-20 | fail→fail | 15,640 | 16,596 | +6% | 1 | 1 | 0% | 3,086 | 3,571 | +16% | 0 | 0 | — |
case-21 | fail→fail | 11,535 | 8,497 | -26% | 1 | 1 | 0% | 2,180 | 2,115 | -3% | 0 | 0 | — |
case-22 | fail→fail | 15,256 | 12,484 | -18% | 1 | 1 | 0% | 2,846 | 2,984 | +5% | 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, and 20 counted toward the lift figure. The other 2 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 +32 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.