@lythos/skill-arena 0.15.7 → 0.16.1

package/src/comparative-judge.test.ts

@@ -1,10 +1,10 @@
-import { describe, test, expect } from 'bun:test'
+import { describe, it, expect } from 'bun:test'
 import { computePareto, buildComparativePrompt, toScoreMatrix, normalizeComparativeOutput } from './comparative-judge'
 import { ArenaManifest, CriterionDef, ComparativeReport } from '@lythos/test-utils/schema'
 import type { ArenaManifest as ArenaManifestType } from '@lythos/test-utils/schema'
 
 describe('computePareto', () => {
-  test('single participant is always non-dominated', () => {
+  it('single participant is always non-dominated', () => {
     const result = computePareto([
       { participant_id: 'run-01', scores: { a: 5, b: 3 } },
     ])
@@ -13,7 +13,7 @@ describe('computePareto', () => {
     expect(result[0].dominated_by).toEqual([])
   })
 
-  test('clear dominance: run-01 dominates run-02 on all criteria', () => {
+  it('clear dominance: run-01 dominates run-02 on all criteria', () => {
     const result = computePareto([
       { participant_id: 'run-01', scores: { coverage: 5, relevance: 5 } },
       { participant_id: 'run-02', scores: { coverage: 3, relevance: 2 } },
@@ -23,7 +23,7 @@ describe('computePareto', () => {
     expect(result[1].dominated_by).toEqual(['run-01'])
   })
 
-  test('equal scores: no one dominates', () => {
+  it('equal scores: no one dominates', () => {
     const result = computePareto([
       { participant_id: 'run-01', scores: { a: 4, b: 4 } },
       { participant_id: 'run-02', scores: { a: 4, b: 4 } },
@@ -32,7 +32,7 @@ describe('computePareto', () => {
     expect(result[1].dominated).toBe(false)
   })
 
-  test('cross dominance: each wins on different criteria', () => {
+  it('cross dominance: each wins on different criteria', () => {
     const result = computePareto([
       { participant_id: 'run-01', scores: { speed: 5, accuracy: 2 } },
       { participant_id: 'run-02', scores: { speed: 2, accuracy: 5 } },
@@ -42,7 +42,7 @@ describe('computePareto', () => {
     expect(result[1].dominated).toBe(false)
   })
 
-  test('multi-participant: transitive dominance chain', () => {
+  it('multi-participant: transitive dominance chain', () => {
     const result = computePareto([
       { participant_id: 'best', scores: { a: 5, b: 5, c: 5 } },
       { participant_id: 'mid', scores: { a: 4, b: 4, c: 4 } },
@@ -56,7 +56,7 @@ describe('computePareto', () => {
     expect(result[2].dominated_by.sort()).toEqual(['best', 'mid'].sort())
   })
 
-  test('Pareto frontier from playground BDD-research: run-01 dominates run-02', () => {
+  it('Pareto frontier from playground BDD-research: run-01 dominates run-02', () => {
     // From playground/arena-bdd-research/report.md:
     // Run-01: coverage=5, relevance=5, actionability=5, depth=5
     // Run-02: coverage=3, relevance=2, actionability=2, depth=1
@@ -69,7 +69,7 @@ describe('computePareto', () => {
     expect(result[1].dominated_by).toEqual(['run-01'])
   })
 
-  test('empty scores object', () => {
+  it('empty scores object', () => {
     const result = computePareto([
       { participant_id: 'a', scores: {} },
       { participant_id: 'b', scores: {} },
@@ -79,7 +79,7 @@ describe('computePareto', () => {
     expect(result[1].dominated).toBe(false)
   })
 
-  test('partial criteria overlap', () => {
+  it('partial criteria overlap', () => {
     const result = computePareto([
       { participant_id: 'run-01', scores: { a: 5, b: 3 } },
       { participant_id: 'run-02', scores: { a: 3, c: 5 } },
@@ -103,7 +103,7 @@ const manifestFixture: ArenaManifest = {
 }
 
 describe('buildComparativePrompt', () => {
-  test('includes task description', () => {
+  it('includes task description', () => {
     const prompt = buildComparativePrompt({
       manifest: manifestFixture,
       verdicts: [],
@@ -111,7 +111,7 @@ describe('buildComparativePrompt', () => {
     expect(prompt).toContain('Write a function that adds two numbers')
   })
 
-  test('includes all participants', () => {
+  it('includes all participants', () => {
     const prompt = buildComparativePrompt({
       manifest: manifestFixture,
       verdicts: [],
@@ -122,7 +122,7 @@ describe('buildComparativePrompt', () => {
     expect(prompt).toContain('Full test discipline')
   })
 
-  test('includes criteria list', () => {
+  it('includes criteria list', () => {
     const prompt = buildComparativePrompt({
       manifest: manifestFixture,
       verdicts: [],
@@ -131,7 +131,7 @@ describe('buildComparativePrompt', () => {
     expect(prompt).toContain('efficiency')
   })
 
-  test('includes Zod schema in output spec', () => {
+  it('includes Zod schema in output spec', () => {
     const prompt = buildComparativePrompt({
       manifest: manifestFixture,
       verdicts: [],
@@ -145,7 +145,7 @@ describe('buildComparativePrompt', () => {
 // ── toScoreMatrix (pure Zod validation wrapper) ──────────────────────
 
 describe('toScoreMatrix', () => {
-  test('passes through valid score cells', () => {
+  it('passes through valid score cells', () => {
     const result = toScoreMatrix(manifestFixture, [
       { participant_id: 'bare', criterion: 'correctness', weight: 0.5, score: 4, rationale: 'works' },
       { participant_id: 'bare', criterion: 'efficiency', weight: 0.5, score: 3, rationale: 'ok' },
@@ -166,7 +166,7 @@ const sampleScoreMatrix = [
 ]
 
 describe('normalizeComparativeOutput', () => {
-  test('passes through already-correct format', () => {
+  it('passes through already-correct format', () => {
     const input = {
       score_matrix: sampleScoreMatrix,
       key_findings: ['TDD produced cleaner code'],
@@ -177,7 +177,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].participant_id).toBe('bare')
   })
 
-  test('maps participantId to participant_id', () => {
+  it('maps participantId to participant_id', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participantId: 'bare', criterion: 'accuracy', weight: 0.5, score: 4, rationale: 'good' },
@@ -189,7 +189,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].participant_id).toBe('bare')
   })
 
-  test('maps side to participant_id', () => {
+  it('maps side to participant_id', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { side: 'tdd', criterion: 'quality', weight: 0.5, score: 5, rationale: 'excellent' },
@@ -201,7 +201,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].participant_id).toBe('tdd')
   })
 
-  test('coerces string score to number', () => {
+  it('coerces string score to number', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', weight: 0.5, score: '4', rationale: 'ok' },
@@ -213,7 +213,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].score).toBe(4)
   })
 
-  test('normalizes weight >1 as percentage', () => {
+  it('normalizes weight >1 as percentage', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', weight: 50, score: 4, rationale: 'ok' },
@@ -225,7 +225,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].weight).toBe(0.5)
   })
 
-  test('defaults weight to 0.25 when undefined', () => {
+  it('defaults weight to 0.25 when undefined', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', score: 4, rationale: 'ok' },
@@ -237,7 +237,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].weight).toBe(0.25)
   })
 
-  test('maps reason to rationale', () => {
+  it('maps reason to rationale', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', weight: 0.5, score: 4, reason: 'looks fine' },
@@ -249,7 +249,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].rationale).toBe('looks fine')
   })
 
-  test('maps explanation to rationale', () => {
+  it('maps explanation to rationale', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', weight: 0.5, score: 4, explanation: 'works' },
@@ -261,7 +261,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.score_matrix as any[])[0].rationale).toBe('works')
   })
 
-  test('normalizes recommendations with role fallback', () => {
+  it('normalizes recommendations with role fallback', () => {
     const input: Record<string, unknown> = {
       score_matrix: [],
       key_findings: [],
@@ -275,7 +275,7 @@ describe('normalizeComparativeOutput', () => {
     expect(recs[0].recommendation).toBe('Use TDD')
   })
 
-  test('normalizes recommendations with advice fallback', () => {
+  it('normalizes recommendations with advice fallback', () => {
     const input: Record<string, unknown> = {
       score_matrix: [],
       key_findings: [],
@@ -287,7 +287,7 @@ describe('normalizeComparativeOutput', () => {
     expect((result.recommendations as any[])[0].recommendation).toBe('Consider refactoring')
   })
 
-  test('handles empty key_findings', () => {
+  it('handles empty key_findings', () => {
     const input: Record<string, unknown> = {
       score_matrix: [],
     }
@@ -295,7 +295,7 @@ describe('normalizeComparativeOutput', () => {
     expect(result.key_findings).toEqual([])
   })
 
-  test('converts pivot-table format: { participant: { criterion: score } }', () => {
+  it('converts pivot-table format: { participant: { criterion: score } }', () => {
     const input: Record<string, unknown> = {
       bare: { correctness: 4, correctness_rationale: 'works', efficiency: 3, efficiency_rationale: 'ok' },
       tdd: { correctness: 5, correctness_rationale: 'tests', efficiency: 4, efficiency_rationale: 'clean' },
@@ -310,7 +310,7 @@ describe('normalizeComparativeOutput', () => {
     expect(bareCorrectness.weight).toBe(0.25)
   })
 
-  test('clamps score to 1-5 range', () => {
+  it('clamps score to 1-5 range', () => {
     const input: Record<string, unknown> = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'a', weight: 0.5, score: 0, rationale: 'terrible' },
@@ -367,7 +367,7 @@ const manifestWithRubrics: ArenaManifestType = {
 }
 
 describe('buildComparativePrompt with structured criteria', () => {
-  test('injects rubric anchors into prompt', () => {
+  it('injects rubric anchors into prompt', () => {
     const prompt = buildComparativePrompt({ manifest: manifestWithRubrics, verdicts: [] })
     expect(prompt).toContain('信息准确性')
     expect(prompt).toContain('Evaluator: ISTJ测试员')
@@ -377,7 +377,7 @@ describe('buildComparativePrompt with structured criteria', () => {
     expect(prompt).toContain('Evaluator: INTJ架构师')
   })
 
-  test('falls back to bare format for string criteria', () => {
+  it('falls back to bare format for string criteria', () => {
     const manifest: ArenaManifestType = {
       id: 'test', created_at: '2026-01-01T00:00:00Z', task: 'test', mode: 'decks',
       participants: [{ id: 'a', name: 'A', deck: 'd1' }, { id: 'b', name: 'B', deck: 'd2' }],
@@ -393,7 +393,7 @@ describe('buildComparativePrompt with structured criteria', () => {
 // Simulate a realistic LLM judge output — the kind of JSON an actual Claude
 // comparative judge call would produce. Verify our normalization handles it.
 describe('full pipeline: mock LLM output → schema validation', () => {
-  test('clean score_matrix passes through ComparativeReport.parse', () => {
+  it('clean score_matrix passes through ComparativeReport.parse', () => {
     const cleanOutput = {
       score_matrix: [
         { participant_id: 'bare', criterion: 'accuracy', weight: 0.4, score: 3, rationale: 'Correct on Bun version but missed pnpm migration detail' },
@@ -419,7 +419,7 @@ describe('full pipeline: mock LLM output → schema validation', () => {
     expect(deepAccuracy!.score).toBe(5)
   })
 
-  test('messy LLM output with field name variants gets normalized', () => {
+  it('messy LLM output with field name variants gets normalized', () => {
     // Simulates a messy Claude output — participantId instead of participant_id,
     // reason instead of rationale, string score
     const messyLLMOutput = {

package/src/player.test.ts

@@ -1,4 +1,4 @@
-import { describe, test, expect } from 'bun:test'
+import { describe, it, expect } from 'bun:test'
 import { resolvePlayer, resolveSides, groupBySide, totalRuns } from './player'
 import { parseArenaToml } from './arena-toml'
 
@@ -20,29 +20,29 @@ deck = "./decks/rich.toml"
 `)
 
 describe('resolvePlayer', () => {
-  test('maps claude-code → claude', () => {
+  it('maps claude-code → claude', () => {
     expect(resolvePlayer('claude-code')).toBe('claude-sdk')
   })
 
-  test('maps Claude → claude (case insensitive)', () => {
+  it('maps Claude → claude (case insensitive)', () => {
     expect(resolvePlayer('Claude')).toBe('claude-sdk')
   })
 
-  test('maps kimi → kimi', () => {
+  it('maps kimi → kimi', () => {
     expect(resolvePlayer('kimi')).toBe('kimi')
   })
 
-  test('passes through unknown player names', () => {
+  it('passes through unknown player names', () => {
     expect(resolvePlayer('expert-architect')).toBe('expert-architect')
   })
 
-  test('trims whitespace', () => {
+  it('trims whitespace', () => {
     expect(resolvePlayer('  claude-code  ')).toBe('claude-sdk')
   })
 })
 
 describe('resolveSides', () => {
-  test('resolves all sides in arena.toml', () => {
+  it('resolves all sides in arena.toml', () => {
     const sides = resolveSides(toml)
     expect(sides).toHaveLength(2)
     expect(sides[0].platform).toBe('claude-sdk')
@@ -50,7 +50,7 @@ describe('resolveSides', () => {
     expect(sides[0].playerName).toBe('claude-code')
   })
 
-  test('preserves side config', () => {
+  it('preserves side config', () => {
     const sides = resolveSides(toml)
     expect(sides[0].side.name).toBe('minimal')
     expect(sides[0].side.deck).toBe('./decks/minimal.toml')
@@ -58,7 +58,7 @@ describe('resolveSides', () => {
 })
 
 describe('groupBySide', () => {
-  test('groups by side name with run count', () => {
+  it('groups by side name with run count', () => {
     const groups = groupBySide(toml)
     expect(groups).toHaveLength(2)
     expect(groups[0].runs).toBe(3) // runs_per_side
@@ -66,7 +66,7 @@ describe('groupBySide', () => {
     expect(groups[0].platform).toBe('claude-sdk')
   })
 
-  test('control flag preserved', () => {
+  it('control flag preserved', () => {
     const controlToml = parseArenaToml(`
 [arena]
 task = "x"
@@ -89,7 +89,7 @@ control = true
 })
 
 describe('totalRuns', () => {
-  test('calculates sides × runs_per_side', () => {
+  it('calculates sides × runs_per_side', () => {
     expect(totalRuns(toml)).toBe(6) // 2 sides × 3 runs
   })
 })