Files
proffesor-for-testing__agen…/tests/fixtures/sample-validation-results.json
Profa 2253d05618 feat: flatten project structure — archive v2, promote v3 to root
Remove the nested v3/ subdirectory by moving all active source code,
tests, configs, and assets directly to the project root. Archive v2/
(dead code, 847MB) to v2-archive.tar.gz.

Key changes:
- Move v3/src/ → src/, v3/tests/ → tests/, v3/packages/ → packages/
- Move v3/tsconfig.json, v3/vitest.config.ts to root
- Merge v3/package.json into root package.json (ESM, scripts, deps, exports)
- Merge v3/scripts/ into scripts/, v3/implementation/ into docs/implementation/
- Update all CI workflows to remove "cd v3 &&" and v3/ path prefixes
- Fix all init installer path resolution (3-level-up → 2-level-up)
- Fix hooks to use "node ./dist/cli/bundle.js" instead of bare "aqe"
- Update settings.json ADR/DDD directories to docs/implementation/
- Fix helper scripts (statusline, adr-compliance, ddd-tracker, guidance-hooks)
- Update 58 agent definitions: aqe/v3/ → aqe/ memory namespaces
- Update 15 guidance shards: v3/src/ → src/ path references
- Fix sync interfaces: remove v3/ and v2/ path references
- Fix test path calculations (process.cwd() + '..' no longer needed)
- Fix flaky perf test threshold (500ms → 1000ms for CI environments)
- Fix complexity-analyzer test to match ADR-051 keyword-based eligibility

Build passes, 17,901 tests pass, CLI and MCP bundles verified.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-01 11:35:41 +00:00

213 lines
6.5 KiB
JSON

[
{
"runId": "run-sonnet-20260202",
"model": "claude-sonnet",
"outcomes": [
{
"skillName": "security-testing",
"trustTier": 3,
"validationLevel": "eval",
"model": "claude-sonnet",
"passed": true,
"score": 0.95,
"testCaseResults": [
{
"testId": "security-injection-001",
"passed": true,
"expectedPatterns": ["sanitize", "validate"],
"actualPatterns": ["sanitize", "validate", "escape"],
"reasoningQuality": 0.92,
"executionTimeMs": 150,
"category": "injection",
"priority": "critical"
},
{
"testId": "security-auth-002",
"passed": true,
"expectedPatterns": ["authenticate", "authorize"],
"actualPatterns": ["authenticate", "authorize"],
"reasoningQuality": 0.88,
"executionTimeMs": 120,
"category": "auth",
"priority": "critical"
},
{
"testId": "security-xss-003",
"passed": true,
"expectedPatterns": ["escape", "sanitize"],
"actualPatterns": ["escape", "sanitize", "encode"],
"reasoningQuality": 0.95,
"executionTimeMs": 100,
"category": "xss",
"priority": "high"
}
],
"timestamp": "2026-02-02T10:00:00.000Z",
"runId": "run-sonnet-20260202",
"metadata": {
"duration": 370,
"environment": "test"
}
},
{
"skillName": "accessibility-testing",
"trustTier": 3,
"validationLevel": "eval",
"model": "claude-sonnet",
"passed": true,
"score": 0.9,
"testCaseResults": [
{
"testId": "a11y-wcag-001",
"passed": true,
"expectedPatterns": ["alt-text", "aria-label"],
"actualPatterns": ["alt-text", "aria-label"],
"reasoningQuality": 0.9,
"executionTimeMs": 200,
"category": "wcag",
"priority": "high"
},
{
"testId": "a11y-keyboard-002",
"passed": true,
"expectedPatterns": ["tab-order", "focus-visible"],
"actualPatterns": ["tab-order", "focus-visible", "keyboard-nav"],
"reasoningQuality": 0.85,
"executionTimeMs": 180,
"category": "keyboard",
"priority": "high"
},
{
"testId": "a11y-color-003",
"passed": false,
"expectedPatterns": ["contrast-ratio"],
"actualPatterns": [],
"reasoningQuality": 0.6,
"executionTimeMs": 150,
"category": "color",
"priority": "medium",
"error": "Contrast ratio check not performed"
}
],
"timestamp": "2026-02-02T10:01:00.000Z",
"runId": "run-sonnet-20260202",
"metadata": {
"duration": 530,
"environment": "test"
}
}
],
"timestamp": "2026-02-02T10:00:00.000Z",
"durationMs": 900,
"metadata": {
"environment": "test",
"version": "1.0.0"
}
},
{
"runId": "run-haiku-20260202",
"model": "claude-haiku",
"outcomes": [
{
"skillName": "security-testing",
"trustTier": 3,
"validationLevel": "eval",
"model": "claude-haiku",
"passed": true,
"score": 0.85,
"testCaseResults": [
{
"testId": "security-injection-001",
"passed": true,
"expectedPatterns": ["sanitize", "validate"],
"actualPatterns": ["sanitize", "validate"],
"reasoningQuality": 0.82,
"executionTimeMs": 80,
"category": "injection",
"priority": "critical"
},
{
"testId": "security-auth-002",
"passed": true,
"expectedPatterns": ["authenticate", "authorize"],
"actualPatterns": ["authenticate"],
"reasoningQuality": 0.75,
"executionTimeMs": 70,
"category": "auth",
"priority": "critical"
},
{
"testId": "security-xss-003",
"passed": false,
"expectedPatterns": ["escape", "sanitize"],
"actualPatterns": ["escape"],
"reasoningQuality": 0.65,
"executionTimeMs": 60,
"category": "xss",
"priority": "high",
"error": "Missing sanitize pattern"
}
],
"timestamp": "2026-02-02T10:00:00.000Z",
"runId": "run-haiku-20260202",
"metadata": {
"duration": 210,
"environment": "test"
}
},
{
"skillName": "accessibility-testing",
"trustTier": 3,
"validationLevel": "eval",
"model": "claude-haiku",
"passed": true,
"score": 0.85,
"testCaseResults": [
{
"testId": "a11y-wcag-001",
"passed": true,
"expectedPatterns": ["alt-text", "aria-label"],
"actualPatterns": ["alt-text", "aria-label"],
"reasoningQuality": 0.85,
"executionTimeMs": 100,
"category": "wcag",
"priority": "high"
},
{
"testId": "a11y-keyboard-002",
"passed": true,
"expectedPatterns": ["tab-order", "focus-visible"],
"actualPatterns": ["tab-order"],
"reasoningQuality": 0.7,
"executionTimeMs": 90,
"category": "keyboard",
"priority": "high"
},
{
"testId": "a11y-color-003",
"passed": true,
"expectedPatterns": ["contrast-ratio"],
"actualPatterns": ["contrast-ratio"],
"reasoningQuality": 0.8,
"executionTimeMs": 80,
"category": "color",
"priority": "medium"
}
],
"timestamp": "2026-02-02T10:01:00.000Z",
"runId": "run-haiku-20260202",
"metadata": {
"duration": 270,
"environment": "test"
}
}
],
"timestamp": "2026-02-02T10:00:00.000Z",
"durationMs": 480,
"metadata": {
"environment": "test",
"version": "1.0.0"
}
}
]