tokengolf 0.3.0 → 0.4.0

package/src/lib/__tests__/score.test.js

@@ -0,0 +1,596 @@
+import { describe, it, expect } from 'vitest';
+import {
+  calculateAchievements,
+  getTier,
+  getEfficiencyRating,
+  getModelClass,
+  MODEL_CLASSES,
+} from '../score.js';
+
+// ── helpers ──────────────────────────────────────────────────────────────────
+
+function keys(run) {
+  return calculateAchievements(run).map((a) => a.key);
+}
+
+function wonRun(overrides = {}) {
+  return {
+    status: 'won',
+    spent: 0.1,
+    budget: 0.5,
+    model: 'claude-sonnet-4-6',
+    promptCount: 5,
+    totalToolCalls: 10,
+    toolCalls: { Read: 5, Edit: 3, Bash: 2 },
+    compactionEvents: [],
+    thinkingInvocations: 0,
+    sessionCount: 1,
+    fainted: false,
+    startedAt: new Date(Date.now() - 15 * 60 * 1000).toISOString(),
+    endedAt: new Date().toISOString(),
+    ...overrides,
+  };
+}
+
+function diedRun(overrides = {}) {
+  return {
+    status: 'died',
+    spent: 0.55,
+    budget: 0.5,
+    model: 'claude-sonnet-4-6',
+    promptCount: 5,
+    totalToolCalls: 10,
+    toolCalls: {},
+    compactionEvents: [],
+    thinkingInvocations: 0,
+    ...overrides,
+  };
+}
+
+// ── pure functions ────────────────────────────────────────────────────────────
+
+describe('getTier', () => {
+  it('Diamond under $0.10', () => expect(getTier(0.05).label).toBe('Diamond'));
+  it('Gold under $0.30', () => expect(getTier(0.2).label).toBe('Gold'));
+  it('Silver under $1.00', () => expect(getTier(0.8).label).toBe('Silver'));
+  it('Bronze under $3.00', () => expect(getTier(2.0).label).toBe('Bronze'));
+  it('Reckless above $3.00', () => expect(getTier(5.0).label).toBe('Reckless'));
+});
+
+describe('getEfficiencyRating', () => {
+  it('LEGENDARY at 24%', () => expect(getEfficiencyRating(0.24, 1.0).label).toBe('LEGENDARY'));
+  it('EFFICIENT at 49%', () => expect(getEfficiencyRating(0.49, 1.0).label).toBe('EFFICIENT'));
+  it('SOLID at 74%', () => expect(getEfficiencyRating(0.74, 1.0).label).toBe('SOLID'));
+  it('CLOSE CALL at 99%', () => expect(getEfficiencyRating(0.99, 1.0).label).toBe('CLOSE CALL'));
+  it('BUSTED over 100%', () => expect(getEfficiencyRating(1.01, 1.0).label).toBe('BUSTED'));
+});
+
+describe('getModelClass', () => {
+  it('haiku', () => expect(getModelClass('claude-haiku-4-5')).toBe(MODEL_CLASSES.haiku));
+  it('sonnet', () => expect(getModelClass('claude-sonnet-4-6')).toBe(MODEL_CLASSES.sonnet));
+  it('opus', () => expect(getModelClass('claude-opus-4-6')).toBe(MODEL_CLASSES.opus));
+  it('unknown defaults to sonnet', () =>
+    expect(getModelClass('unknown-model')).toBe(MODEL_CLASSES.sonnet));
+});
+
+// ── model class achievements ──────────────────────────────────────────────────
+
+describe('model class achievements', () => {
+  it('haiku win → gold_haiku', () => {
+    expect(keys(wonRun({ model: 'claude-haiku-4-5-20251001' }))).toContain('gold_haiku');
+  });
+  it('haiku win under $0.10 → diamond', () => {
+    expect(keys(wonRun({ model: 'claude-haiku-4-5-20251001', spent: 0.08 }))).toContain('diamond');
+  });
+  it('haiku win at $0.12 → no diamond', () => {
+    expect(keys(wonRun({ model: 'claude-haiku-4-5-20251001', spent: 0.12 }))).not.toContain(
+      'diamond'
+    );
+  });
+  it('sonnet win → silver_sonnet', () => {
+    expect(keys(wonRun())).toContain('silver_sonnet');
+  });
+  it('opus win → bronze_opus', () => {
+    expect(keys(wonRun({ model: 'claude-opus-4-6' }))).toContain('bronze_opus');
+  });
+});
+
+// ── budget efficiency achievements ───────────────────────────────────────────
+
+describe('budget efficiency', () => {
+  it('sniper at 20% budget', () => {
+    expect(keys(wonRun({ spent: 0.1, budget: 0.5 }))).toContain('sniper');
+  });
+  it('efficient at 40% budget', () => {
+    expect(keys(wonRun({ spent: 0.2, budget: 0.5 }))).toContain('efficient');
+  });
+  it('no sniper at 30% budget', () => {
+    expect(keys(wonRun({ spent: 0.15, budget: 0.5 }))).not.toContain('sniper');
+  });
+  it('penny pincher under $0.10', () => {
+    expect(keys(wonRun({ spent: 0.08 }))).toContain('penny');
+  });
+  it('no penny at $0.12', () => {
+    expect(keys(wonRun({ spent: 0.12 }))).not.toContain('penny');
+  });
+  it('no sniper/efficient without budget', () => {
+    const a = keys(wonRun({ budget: null, spent: 0.05 }));
+    expect(a).not.toContain('sniper');
+    expect(a).not.toContain('efficient');
+  });
+});
+
+// ── prompting skill ───────────────────────────────────────────────────────────
+
+describe('prompting skill achievements', () => {
+  it('one_shot on single prompt', () => {
+    expect(keys(wonRun({ promptCount: 1 }))).toContain('one_shot');
+  });
+  it('no one_shot on 2 prompts', () => {
+    expect(keys(wonRun({ promptCount: 2 }))).not.toContain('one_shot');
+  });
+  it('conversationalist at 20+ prompts', () => {
+    expect(keys(wonRun({ promptCount: 20 }))).toContain('conversationalist');
+  });
+  it('terse: ≤3 prompts and ≥10 tool calls', () => {
+    expect(keys(wonRun({ promptCount: 2, totalToolCalls: 12 }))).toContain('terse');
+  });
+  it('no terse if tool calls < 10', () => {
+    expect(keys(wonRun({ promptCount: 2, totalToolCalls: 8 }))).not.toContain('terse');
+  });
+  it('high_leverage: 5+ tools per prompt', () => {
+    expect(keys(wonRun({ promptCount: 2, totalToolCalls: 10 }))).toContain('high_leverage');
+  });
+  it('backseat_driver: 15+ prompts, <1 tool per prompt', () => {
+    expect(keys(wonRun({ promptCount: 15, totalToolCalls: 10 }))).toContain('backseat_driver');
+  });
+});
+
+// ── tool mastery ──────────────────────────────────────────────────────────────
+
+describe('tool mastery achievements', () => {
+  it('read_only: reads but no edits/writes', () => {
+    expect(keys(wonRun({ toolCalls: { Read: 5 }, totalToolCalls: 5 }))).toContain('read_only');
+  });
+  it('no read_only if Edit present', () => {
+    expect(keys(wonRun({ toolCalls: { Read: 5, Edit: 1 }, totalToolCalls: 6 }))).not.toContain(
+      'read_only'
+    );
+  });
+  it('editor: 10+ Edit calls', () => {
+    expect(keys(wonRun({ toolCalls: { Edit: 10 }, totalToolCalls: 10 }))).toContain('editor');
+  });
+  it('bash_warrior: 10+ Bash, ≥50% of tools', () => {
+    expect(keys(wonRun({ toolCalls: { Bash: 10, Read: 5 }, totalToolCalls: 15 }))).toContain(
+      'bash_warrior'
+    ); // 67%
+    expect(keys(wonRun({ toolCalls: { Bash: 10, Read: 9 }, totalToolCalls: 19 }))).toContain(
+      'bash_warrior'
+    ); // 53%
+    expect(keys(wonRun({ toolCalls: { Bash: 10 }, totalToolCalls: 10 }))).toContain('bash_warrior'); // 100%
+    expect(keys(wonRun({ toolCalls: { Bash: 5, Read: 6 }, totalToolCalls: 11 }))).not.toContain(
+      'bash_warrior'
+    ); // <10 Bash
+  });
+  it('scout: ≥60% Read, ≥5 total', () => {
+    expect(keys(wonRun({ toolCalls: { Read: 7, Edit: 3 }, totalToolCalls: 10 }))).toContain(
+      'scout'
+    );
+  });
+  it('no scout if <60% Read', () => {
+    expect(keys(wonRun({ toolCalls: { Read: 5, Edit: 5 }, totalToolCalls: 10 }))).not.toContain(
+      'scout'
+    );
+  });
+  it('surgeon: 1-3 Edits, under budget', () => {
+    expect(
+      keys(wonRun({ toolCalls: { Read: 5, Edit: 2 }, totalToolCalls: 7, spent: 0.1, budget: 0.5 }))
+    ).toContain('surgeon');
+  });
+  it('no surgeon if over budget', () => {
+    expect(
+      keys(wonRun({ toolCalls: { Edit: 2 }, totalToolCalls: 2, spent: 0.6, budget: 0.5 }))
+    ).not.toContain('surgeon');
+  });
+  it('toolbox: 5+ distinct tools', () => {
+    expect(
+      keys(
+        wonRun({ toolCalls: { Read: 1, Edit: 1, Bash: 1, Glob: 1, Grep: 1 }, totalToolCalls: 5 })
+      )
+    ).toContain('toolbox');
+  });
+});
+
+// ── cost per prompt ───────────────────────────────────────────────────────────
+
+describe('cost per prompt', () => {
+  it('cheap_shots: <$0.01 per prompt, ≥3 prompts', () => {
+    expect(keys(wonRun({ spent: 0.02, promptCount: 3 }))).toContain('cheap_shots');
+  });
+  it('expensive_taste on won run: ≥$0.50 per prompt, ≥3 prompts', () => {
+    expect(keys(wonRun({ spent: 1.6, promptCount: 3, budget: 2.0 }))).toContain('expensive_taste');
+  });
+});
+
+// ── time-based ────────────────────────────────────────────────────────────────
+
+describe('time-based achievements', () => {
+  it('speedrun under 5 minutes', () => {
+    const start = new Date(Date.now() - 3 * 60 * 1000).toISOString();
+    expect(keys(wonRun({ startedAt: start, endedAt: new Date().toISOString() }))).toContain(
+      'speedrun'
+    );
+  });
+  it('no speedrun at 10 minutes', () => {
+    const start = new Date(Date.now() - 10 * 60 * 1000).toISOString();
+    expect(keys(wonRun({ startedAt: start, endedAt: new Date().toISOString() }))).not.toContain(
+      'speedrun'
+    );
+  });
+  it('marathon over 60 minutes', () => {
+    const start = new Date(Date.now() - 90 * 60 * 1000).toISOString();
+    expect(keys(wonRun({ startedAt: start, endedAt: new Date().toISOString() }))).toContain(
+      'marathon'
+    );
+  });
+  it('endurance over 3 hours', () => {
+    const start = new Date(Date.now() - 4 * 60 * 60 * 1000).toISOString();
+    expect(keys(wonRun({ startedAt: start, endedAt: new Date().toISOString() }))).toContain(
+      'endurance'
+    );
+  });
+  it('endurance but not marathon at 4 hours', () => {
+    const start = new Date(Date.now() - 4 * 60 * 60 * 1000).toISOString();
+    const a = keys(wonRun({ startedAt: start, endedAt: new Date().toISOString() }));
+    expect(a).toContain('endurance');
+    expect(a).not.toContain('marathon');
+  });
+});
+
+// ── death marks ───────────────────────────────────────────────────────────────
+
+describe('death marks', () => {
+  it('blowout at 2× budget', () => {
+    expect(keys(diedRun({ spent: 1.0, budget: 0.5 }))).toContain('blowout');
+  });
+  it('no blowout at 1.5×', () => {
+    expect(keys(diedRun({ spent: 0.75, budget: 0.5 }))).not.toContain('blowout');
+  });
+  it('so_close between 100-110% budget', () => {
+    expect(keys(diedRun({ spent: 0.52, budget: 0.5 }))).toContain('so_close');
+  });
+  it('no so_close at 115%', () => {
+    expect(keys(diedRun({ spent: 0.575, budget: 0.5 }))).not.toContain('so_close');
+  });
+  it('tool_happy at 30+ tool calls', () => {
+    expect(keys(diedRun({ totalToolCalls: 30 }))).toContain('tool_happy');
+  });
+  it('no tool_happy at 29', () => {
+    expect(keys(diedRun({ totalToolCalls: 29 }))).not.toContain('tool_happy');
+  });
+  it('silent_death at ≤2 prompts', () => {
+    expect(keys(diedRun({ promptCount: 2 }))).toContain('silent_death');
+  });
+  it('no silent_death at 3 prompts', () => {
+    expect(keys(diedRun({ promptCount: 3 }))).not.toContain('silent_death');
+  });
+  it('fumble at 5+ failed tool calls on death', () => {
+    expect(keys(diedRun({ failedToolCalls: 5 }))).toContain('fumble');
+  });
+});
+
+// ── death marks don't fire on won runs ───────────────────────────────────────
+
+describe("death marks don't fire on won runs", () => {
+  it('no blowout on won', () => {
+    expect(keys(wonRun({ spent: 2.0, budget: 0.5 }))).not.toContain('blowout');
+  });
+  it('no tool_happy on won', () => {
+    expect(keys(wonRun({ totalToolCalls: 35 }))).not.toContain('tool_happy');
+  });
+  it('no silent_death on won', () => {
+    expect(keys(wonRun({ promptCount: 1 }))).not.toContain('silent_death');
+  });
+});
+
+// ── hubris / ultrathink ───────────────────────────────────────────────────────
+
+describe('hubris + ultrathink', () => {
+  it('hubris fires on death with thinking', () => {
+    expect(keys(diedRun({ thinkingInvocations: 2 }))).toContain('hubris');
+  });
+  it('no hubris on won with thinking', () => {
+    expect(keys(wonRun({ thinkingInvocations: 2 }))).not.toContain('hubris');
+  });
+  it('spell_cast on won with thinking', () => {
+    expect(keys(wonRun({ thinkingInvocations: 1 }))).toContain('spell_cast');
+  });
+  it('deep_thinker at 3+ invocations', () => {
+    expect(keys(wonRun({ thinkingInvocations: 3 }))).toContain('deep_thinker');
+  });
+  it('silent_run on won with 0 thinking and SOLID budget', () => {
+    expect(keys(wonRun({ thinkingInvocations: 0, spent: 0.3, budget: 0.5 }))).toContain(
+      'silent_run'
+    );
+  });
+  it('no silent_run when thinking undefined', () => {
+    const run = wonRun();
+    delete run.thinkingInvocations;
+    expect(keys(run)).not.toContain('silent_run');
+  });
+});
+
+// ── phase 2: new hook fields ──────────────────────────────────────────────────
+
+describe('failed tool call achievements', () => {
+  it('clean_run: 0 failures, ≥5 tool calls', () => {
+    expect(keys(wonRun({ failedToolCalls: 0, totalToolCalls: 10 }))).toContain('clean_run');
+  });
+  it('no clean_run if <5 tool calls', () => {
+    expect(keys(wonRun({ failedToolCalls: 0, totalToolCalls: 4 }))).not.toContain('clean_run');
+  });
+  it('no clean_run if failedToolCalls undefined but totalToolCalls ok', () => {
+    // undefined ?? 0 = 0 → should still fire
+    const run = wonRun({ totalToolCalls: 10 });
+    delete run.failedToolCalls;
+    expect(keys(run)).toContain('clean_run');
+  });
+  it('stubborn: 10+ failures, still won', () => {
+    expect(keys(wonRun({ failedToolCalls: 10 }))).toContain('stubborn');
+  });
+});
+
+describe('subagent achievements', () => {
+  it('lone_wolf: 0 subagents', () => {
+    expect(keys(wonRun({ subagentSpawns: 0 }))).toContain('lone_wolf');
+  });
+  it('no lone_wolf if subagents undefined (defaults to 0)', () => {
+    const run = wonRun();
+    delete run.subagentSpawns;
+    expect(keys(run)).toContain('lone_wolf');
+  });
+  it('summoner: 5+ subagents', () => {
+    expect(keys(wonRun({ subagentSpawns: 5 }))).toContain('summoner');
+  });
+  it('army: 10+ subagents, <50% budget', () => {
+    expect(keys(wonRun({ subagentSpawns: 10, spent: 0.1, budget: 0.5 }))).toContain('army');
+  });
+  it('no army if ≥50% budget', () => {
+    expect(keys(wonRun({ subagentSpawns: 10, spent: 0.3, budget: 0.5 }))).not.toContain('army');
+  });
+});
+
+describe('turn count achievements', () => {
+  it('agentic: 3+ turns per prompt', () => {
+    expect(keys(wonRun({ turnCount: 15, promptCount: 5 }))).toContain('agentic');
+  });
+  it('no agentic below 3× ratio', () => {
+    expect(keys(wonRun({ turnCount: 10, promptCount: 5 }))).not.toContain('agentic');
+  });
+  it('obedient: turnCount === promptCount, ≥3 prompts', () => {
+    expect(keys(wonRun({ turnCount: 5, promptCount: 5 }))).toContain('obedient');
+  });
+  it('no obedient if mismatch', () => {
+    expect(keys(wonRun({ turnCount: 6, promptCount: 5 }))).not.toContain('obedient');
+  });
+  it('no obedient if <3 prompts', () => {
+    expect(keys(wonRun({ turnCount: 2, promptCount: 2 }))).not.toContain('obedient');
+  });
+});
+
+// ── session / rest achievements ───────────────────────────────────────────────
+
+describe('session achievements', () => {
+  it('no_rest on single session', () => {
+    expect(keys(wonRun({ sessionCount: 1 }))).toContain('no_rest');
+  });
+  it('made_camp on 2+ sessions', () => {
+    expect(keys(wonRun({ sessionCount: 2 }))).toContain('made_camp');
+  });
+  it('came_back if fainted', () => {
+    expect(keys(wonRun({ fainted: true, sessionCount: 2 }))).toContain('came_back');
+  });
+});
+
+// ── getModelClass opusplan ────────────────────────────────────────────────────
+
+describe('getModelClass opusplan', () => {
+  it('opusplan string → MODEL_CLASSES.opusplan', () => {
+    expect(getModelClass('opusplan')).toBe(MODEL_CLASSES.opusplan);
+  });
+});
+
+// ── Paladin achievements ──────────────────────────────────────────────────────
+
+describe('Paladin achievements', () => {
+  it('paladin: opusplan win', () => {
+    expect(keys(wonRun({ model: 'opusplan' }))).toContain('paladin');
+  });
+  it('grand_strategist: opusplan win at ≤25% budget', () => {
+    expect(keys(wonRun({ model: 'opusplan', spent: 0.1, budget: 0.5 }))).toContain(
+      'grand_strategist'
+    );
+  });
+  it('no grand_strategist at 30% budget', () => {
+    expect(keys(wonRun({ model: 'opusplan', spent: 0.15, budget: 0.5 }))).not.toContain(
+      'grand_strategist'
+    );
+  });
+  it('architect: opus pct > 60%', () => {
+    const mb = { 'claude-opus-4-6': 0.07, 'claude-sonnet-4-6': 0.03 };
+    expect(keys(wonRun({ model: 'opusplan', spent: 0.1, modelBreakdown: mb }))).toContain(
+      'architect'
+    );
+  });
+  it('blitz: opus pct < 25%', () => {
+    const mb = { 'claude-opus-4-6': 0.02, 'claude-sonnet-4-6': 0.08 };
+    expect(keys(wonRun({ model: 'opusplan', spent: 0.1, modelBreakdown: mb }))).toContain('blitz');
+  });
+  it('equilibrium: opus pct 40–60%', () => {
+    const mb = { 'claude-opus-4-6': 0.05, 'claude-sonnet-4-6': 0.05 };
+    expect(keys(wonRun({ model: 'opusplan', spent: 0.1, modelBreakdown: mb }))).toContain(
+      'equilibrium'
+    );
+  });
+  it('Paladin does not fire purist/committed/chameleon', () => {
+    const a = keys(wonRun({ model: 'opusplan', spent: 0.1, budget: 0.5 }));
+    expect(a).not.toContain('purist');
+    expect(a).not.toContain('committed');
+    expect(a).not.toContain('chameleon');
+  });
+});
+
+// ── effort-based achievements ─────────────────────────────────────────────────
+
+describe('effort-based achievements', () => {
+  it('speedrunner: low effort, completed under budget', () => {
+    expect(keys(wonRun({ effort: 'low', spent: 0.1, budget: 0.5 }))).toContain('speedrunner');
+  });
+  it('no speedrunner if over budget', () => {
+    expect(keys(wonRun({ effort: 'low', spent: 0.6, budget: 0.5 }))).not.toContain('speedrunner');
+  });
+  it('tryhard: high effort, ≤25% budget', () => {
+    expect(keys(wonRun({ effort: 'high', spent: 0.1, budget: 0.5 }))).toContain('tryhard');
+  });
+  it('no tryhard at 30%', () => {
+    expect(keys(wonRun({ effort: 'high', spent: 0.15, budget: 0.5 }))).not.toContain('tryhard');
+  });
+  it('archmagus: max effort, opus model', () => {
+    expect(keys(wonRun({ effort: 'max', model: 'claude-opus-4-6' }))).toContain('archmagus');
+  });
+  it('no archmagus: max effort, non-opus', () => {
+    expect(keys(wonRun({ effort: 'max', model: 'claude-sonnet-4-6' }))).not.toContain('archmagus');
+  });
+});
+
+// ── fast mode achievements ────────────────────────────────────────────────────
+
+describe('fast mode achievements', () => {
+  it('lightning: opus fastMode, under budget', () => {
+    expect(
+      keys(wonRun({ fastMode: true, model: 'claude-opus-4-6', spent: 0.1, budget: 0.5 }))
+    ).toContain('lightning');
+  });
+  it('daredevil: opus fastMode, ≤25% budget', () => {
+    expect(
+      keys(wonRun({ fastMode: true, model: 'claude-opus-4-6', spent: 0.1, budget: 0.5 }))
+    ).toContain('daredevil');
+  });
+  it('no lightning for non-opus fastMode', () => {
+    expect(
+      keys(wonRun({ fastMode: true, model: 'claude-sonnet-4-6', spent: 0.1, budget: 0.5 }))
+    ).not.toContain('lightning');
+  });
+});
+
+// ── compaction / gear achievements ────────────────────────────────────────────
+
+describe('compaction achievements', () => {
+  it('overencumbered: auto-compaction event', () => {
+    const events = [{ trigger: 'auto', contextPct: 92 }];
+    expect(keys(wonRun({ compactionEvents: events }))).toContain('overencumbered');
+  });
+  it('no overencumbered with only manual compaction', () => {
+    const events = [{ trigger: 'manual', contextPct: 45 }];
+    expect(keys(wonRun({ compactionEvents: events }))).not.toContain('overencumbered');
+  });
+  it('ghost_run: manual compact at ≤30% context', () => {
+    const events = [{ trigger: 'manual', contextPct: 28 }];
+    expect(keys(wonRun({ compactionEvents: events }))).toContain('ghost_run');
+  });
+  it('ultralight: manual compact at 31–40% context', () => {
+    const events = [{ trigger: 'manual', contextPct: 35 }];
+    expect(keys(wonRun({ compactionEvents: events }))).toContain('ultralight');
+  });
+  it('traveling_light: manual compact at 41–50% context', () => {
+    const events = [{ trigger: 'manual', contextPct: 48 }];
+    expect(keys(wonRun({ compactionEvents: events }))).toContain('traveling_light');
+  });
+  it('no compaction achievement above 50%', () => {
+    const events = [{ trigger: 'manual', contextPct: 60 }];
+    const a = keys(wonRun({ compactionEvents: events }));
+    expect(a).not.toContain('ghost_run');
+    expect(a).not.toContain('ultralight');
+    expect(a).not.toContain('traveling_light');
+  });
+});
+
+// ── calculated_risk ───────────────────────────────────────────────────────────
+
+describe('calculated_risk', () => {
+  it('fires with thinking + ≤25% budget', () => {
+    expect(keys(wonRun({ thinkingInvocations: 1, spent: 0.1, budget: 0.5 }))).toContain(
+      'calculated_risk'
+    );
+  });
+  it('no calculated_risk at 30% budget', () => {
+    expect(keys(wonRun({ thinkingInvocations: 1, spent: 0.15, budget: 0.5 }))).not.toContain(
+      'calculated_risk'
+    );
+  });
+});
+
+// ── multi-model achievements ──────────────────────────────────────────────────
+
+describe('frugal + rogue_run', () => {
+  it('frugal: haiku ≥50% of spend', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.06, 'claude-sonnet-4-6': 0.04 };
+    expect(keys(wonRun({ modelBreakdown: mb, spent: 0.1 }))).toContain('frugal');
+  });
+  it('rogue_run: haiku ≥75% of spend', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.08, 'claude-sonnet-4-6': 0.02 };
+    expect(keys(wonRun({ modelBreakdown: mb, spent: 0.1 }))).toContain('rogue_run');
+  });
+  it('no frugal when haiku < 50%', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.04, 'claude-sonnet-4-6': 0.06 };
+    expect(keys(wonRun({ modelBreakdown: mb, spent: 0.1 }))).not.toContain('frugal');
+  });
+  it('no rogue_run when haiku < 75%', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.06, 'claude-sonnet-4-6': 0.04 };
+    expect(keys(wonRun({ modelBreakdown: mb, spent: 0.1 }))).not.toContain('rogue_run');
+  });
+});
+
+// ── model switching achievements ──────────────────────────────────────────────
+
+describe('model switching achievements', () => {
+  it('purist: single distinct model', () => {
+    expect(keys(wonRun({ modelSwitches: 0, distinctModels: 1 }))).toContain('purist');
+  });
+  it('committed: 0 switches, ≤1 distinct', () => {
+    expect(keys(wonRun({ modelSwitches: 0, distinctModels: 1 }))).toContain('committed');
+  });
+  it('chameleon: 2+ distinct models, under budget', () => {
+    expect(
+      keys(wonRun({ modelSwitches: 2, distinctModels: 2, spent: 0.1, budget: 0.5 }))
+    ).toContain('chameleon');
+  });
+  it('no chameleon if over budget', () => {
+    expect(
+      keys(wonRun({ modelSwitches: 2, distinctModels: 2, spent: 0.6, budget: 0.5 }))
+    ).not.toContain('chameleon');
+  });
+  it('tactical_switch: exactly 1 switch, under budget', () => {
+    expect(
+      keys(wonRun({ modelSwitches: 1, distinctModels: 2, spent: 0.1, budget: 0.5 }))
+    ).toContain('tactical_switch');
+  });
+  it('class_defection: declared haiku but >50% on heavier models', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.04, 'claude-sonnet-4-6': 0.06 };
+    expect(
+      keys(wonRun({ model: 'claude-haiku-4-5-20251001', modelBreakdown: mb, spent: 0.1 }))
+    ).toContain('class_defection');
+  });
+  it('no class_defection: declared haiku, haiku dominant', () => {
+    const mb = { 'claude-haiku-4-5-20251001': 0.07, 'claude-sonnet-4-6': 0.03 };
+    expect(
+      keys(wonRun({ model: 'claude-haiku-4-5-20251001', modelBreakdown: mb, spent: 0.1 }))
+    ).not.toContain('class_defection');
+  });
+  it('class_defection: declared sonnet but >40% on opus', () => {
+    const mb = { 'claude-sonnet-4-6': 0.05, 'claude-opus-4-6': 0.05 };
+    expect(keys(wonRun({ model: 'claude-sonnet-4-6', modelBreakdown: mb, spent: 0.1 }))).toContain(
+      'class_defection'
+    );
+  });
+});