voyage/tests/validators/profile-validator.test.mjs
Kjell Tore Guttormsen 213cf388de chore(voyage): S34 — V30 economy-profile self-declares experimental (uncalibrated Jaccard floor)
The economy profile's cross-tier Jaccard floor (0.55) rests on parked-synthetic
fixtures; empirical Step-17 calibration is v4.2-budget-gated ($60–120,
unauthorized). Fork resolved as label-not-calibrate: the experimental status,
previously prose-only in docs/profiles.md, now lives in the profile data and is
machine-checked. No new user-facing capability — honest labeling + a guard.

- lib/profiles/economy.yaml: add `experimental: true` (with rationale comment).
- lib/validators/profile-validator.mjs: recognize `experimental` as an OPTIONAL
  boolean; non-boolean → PROFILE_INVALID_ENUM. Absent ⇒ tier is stable
  (premium/balanced unaffected, profile_version stays 1.0 — additive).
- README.md + docs/operations.md + docs/profiles.md: flag the `economy` table
  row "⚠ Experimental (uncalibrated Jaccard floor)".
- tests/synthetic/profile-jaccard-calibration.md + analysis §6/§10 + backlog
  plan §S34: cross-reference the marker; mark V30/S34 done.

+5 tests (739 → 744, 742/2/0): economy declares experimental:true; premium and
balanced do not; validator rejects non-boolean experimental; every profile-doc
economy row is flagged; the flag tracks the calibration's parked-synthetic
status (must drop in the same change that lands real calibration).

Closes the balance backlog (4/4, S31–S34). claude plugin validate green.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LqBYc8Ltrk7LipyJmGxXiB
2026-06-20 10:18:38 +02:00

204 lines
7.5 KiB
JavaScript

// tests/validators/profile-validator.test.mjs
// SC #1, #2, #3: profile-validator validates lib/profiles/{economy,balanced,premium}.yaml
// (innebygde profiler) plus rejects invalid models and invalid enum types.
import { test } from 'node:test';
import { strict as assert } from 'node:assert';
import { join } from 'node:path';
import { fileURLToPath } from 'node:url';
import { dirname } from 'node:path';
import {
validateProfile,
validateProfileContent,
PROFILE_REQUIRED_FIELDS,
PROFILE_REQUIRED_PHASES,
} from '../../lib/validators/profile-validator.mjs';
const __dirname = dirname(fileURLToPath(import.meta.url));
const REPO_ROOT = join(__dirname, '..', '..');
// SC #1: alle 3 innebygde profiler grønne
for (const profileName of ['economy', 'balanced', 'premium']) {
test(`SC #1: lib/profiles/${profileName}.yaml validates clean`, () => {
const r = validateProfile(join(REPO_ROOT, 'lib', 'profiles', `${profileName}.yaml`));
assert.equal(r.valid, true,
`expected valid, got errors: ${JSON.stringify(r.errors, null, 2)}`);
assert.equal(r.errors.length, 0);
// Spot-check: all 6 phases present
const phases = r.parsed.frontmatter.phase_models.map(p => p.phase);
for (const required of PROFILE_REQUIRED_PHASES) {
assert.ok(phases.includes(required), `${profileName} missing phase: ${required}`);
}
});
}
// SC #2: PROFILE_INVALID_MODEL fired when phase_models[N].model not in allowlist
test('SC #2: profile-invalid-model.yaml rejected with PROFILE_INVALID_MODEL at phase_models[2].model', () => {
const r = validateProfile(join(REPO_ROOT, 'tests', 'fixtures', 'profile-invalid-model.yaml'));
assert.equal(r.valid, false);
const found = r.errors.find(e => e.code === 'PROFILE_INVALID_MODEL');
assert.ok(found, `expected PROFILE_INVALID_MODEL, got: ${JSON.stringify(r.errors)}`);
assert.equal(found.location, 'phase_models[2].model',
`expected location phase_models[2].model, got ${found.location}`);
assert.match(found.message, /gpt-4/);
});
// SC #3: PROFILE_INVALID_ENUM for wrong-type values
test('SC #3: profile-invalid-enum.yaml rejected with PROFILE_INVALID_ENUM (external_research_enabled is string)', () => {
const r = validateProfile(join(REPO_ROOT, 'tests', 'fixtures', 'profile-invalid-enum.yaml'));
assert.equal(r.valid, false);
const found = r.errors.find(e => e.code === 'PROFILE_INVALID_ENUM' && /external_research_enabled/.test(e.message));
assert.ok(found, `expected PROFILE_INVALID_ENUM for external_research_enabled, got: ${JSON.stringify(r.errors)}`);
assert.match(found.message, /boolean/);
});
// VOYAGE_ALLOW_HAIKU env-var allows haiku model
test('VOYAGE_ALLOW_HAIKU=1 allows haiku in phase_models', () => {
const haikuProfile = `---
profile_version: "1.0"
name: haiku-allowed
phase_models:
- phase: brief
model: haiku
- phase: research
model: sonnet
- phase: plan
model: opus
- phase: execute
model: sonnet
- phase: review
model: opus
- phase: continue
model: sonnet
parallel_agents_min: 2
parallel_agents_max: 4
external_research_enabled: false
brief_reviewer_iter_cap: 1
---
`;
// Default env: haiku rejected
const denied = validateProfileContent(haikuProfile, { env: { /* no VOYAGE_ALLOW_HAIKU */ } });
assert.equal(denied.valid, false);
const haikuErr = denied.errors.find(e => e.code === 'PROFILE_INVALID_MODEL' && /haiku/.test(e.message));
assert.ok(haikuErr, `expected haiku rejection: ${JSON.stringify(denied.errors)}`);
assert.match(haikuErr.message, /VOYAGE_ALLOW_HAIKU/);
// With opt-in: haiku accepted
const allowed = validateProfileContent(haikuProfile, { env: { VOYAGE_ALLOW_HAIKU: '1' } });
assert.equal(allowed.valid, true,
`expected valid with VOYAGE_ALLOW_HAIKU=1, got: ${JSON.stringify(allowed.errors)}`);
});
// Required fields presence
test('PROFILE_MISSING_FIELD when name absent', () => {
const r = validateProfileContent(`---
profile_version: "1.0"
phase_models:
- phase: brief
model: sonnet
- phase: research
model: sonnet
- phase: plan
model: opus
- phase: execute
model: sonnet
- phase: review
model: opus
- phase: continue
model: sonnet
parallel_agents_min: 2
parallel_agents_max: 4
external_research_enabled: false
brief_reviewer_iter_cap: 1
---
`);
assert.equal(r.valid, false);
const found = r.errors.find(e => e.code === 'PROFILE_MISSING_FIELD' && /name/.test(e.message));
assert.ok(found, `expected PROFILE_MISSING_FIELD for name, got: ${JSON.stringify(r.errors)}`);
});
// PROFILE_NOT_FOUND for missing file
test('PROFILE_NOT_FOUND for non-existent file', () => {
const r = validateProfile('/tmp/does-not-exist-profile-xyz.yaml');
assert.equal(r.valid, false);
assert.ok(r.errors.find(e => e.code === 'PROFILE_NOT_FOUND'));
});
// REQUIRED_FIELDS frontmatter contract drift-pin
test('PROFILE_REQUIRED_FIELDS export drift-pin', () => {
assert.deepEqual(
[...PROFILE_REQUIRED_FIELDS].sort(),
['brief_reviewer_iter_cap', 'external_research_enabled', 'name',
'parallel_agents_max', 'parallel_agents_min', 'phase_models'].sort(),
'PROFILE_REQUIRED_FIELDS contract drift — pin contract',
);
});
test('PROFILE_REQUIRED_PHASES export drift-pin', () => {
assert.deepEqual(
[...PROFILE_REQUIRED_PHASES].sort(),
['brief', 'research', 'plan', 'execute', 'review', 'continue'].sort(),
'PROFILE_REQUIRED_PHASES contract drift — pin contract',
);
});
// S34 (V30) — economy carries an explicit `experimental` marker.
// The cross-tier Jaccard floor (0.55) rests on parked-synthetic fixtures;
// empirical Step-17 calibration is deferred to v4.2 ($60-120 LLM budget,
// unauthorized). Until a real calibration lands, `economy` is self-declared
// experimental in the profile DATA — not only in prose — so the status
// survives a docs rewrite. See tests/synthetic/profile-jaccard-calibration.md.
test('S34: economy.yaml declares experimental: true (uncalibrated Jaccard floor)', () => {
const r = validateProfile(join(REPO_ROOT, 'lib', 'profiles', 'economy.yaml'));
assert.equal(r.valid, true, `economy.yaml should validate: ${JSON.stringify(r.errors)}`);
assert.equal(r.parsed.frontmatter.experimental, true,
'economy.yaml must carry experimental: true until the Jaccard floor is empirically calibrated (S34/V30)');
});
test('S34: premium + balanced are NOT flagged experimental', () => {
for (const name of ['premium', 'balanced']) {
const r = validateProfile(join(REPO_ROOT, 'lib', 'profiles', `${name}.yaml`));
assert.equal(r.valid, true, `${name}.yaml should validate: ${JSON.stringify(r.errors)}`);
assert.notEqual(r.parsed.frontmatter.experimental, true,
`${name}.yaml must not be flagged experimental`);
}
});
test('S34: validator type-checks experimental as boolean (PROFILE_INVALID_ENUM)', () => {
const bad = `---
profile_version: "1.0"
name: bad-experimental
phase_models:
- phase: brief
model: sonnet
- phase: research
model: sonnet
- phase: plan
model: sonnet
- phase: execute
model: sonnet
- phase: review
model: sonnet
- phase: continue
model: sonnet
parallel_agents_min: 2
parallel_agents_max: 3
external_research_enabled: false
brief_reviewer_iter_cap: 1
experimental: "yes"
---
`;
const r = validateProfileContent(bad);
assert.equal(r.valid, false, 'experimental: "yes" (string) must be rejected');
const found = r.errors.find(e => e.code === 'PROFILE_INVALID_ENUM' && /experimental/.test(e.message));
assert.ok(found, `expected PROFILE_INVALID_ENUM for experimental, got: ${JSON.stringify(r.errors)}`);
assert.match(found.message, /boolean/);
});