TECHNICAL VALIDATION — API & ALGORITHM VERIFICATION

Frequency: Monthly or when APIs change Purpose: Ensure technical foundation is still valid Action: Verify APIs work, algorithms effective

---

INVOCATION

/validate-technical          # Full validation
/validate-technical api      # API endpoints only
/validate-technical algo     # Algorithm effectiveness only

---

VALIDATION PROTOCOL

1. Registry API Validation

PyPI API

# Test package metadata endpoint
curl -s "https://pypi.org/pypi/flask/json" | python -c "
import json, sys
data = json.load(sys.stdin)
print(f'Name: {data[\"info\"][\"name\"]}')
print(f'Version: {data[\"info\"][\"version\"]}')
print(f'Author: {data[\"info\"][\"author\"]}')
print(f'Downloads: Available via stats API')
print(f'Repository: {data[\"info\"].get(\"project_urls\", {}).get(\"Homepage\", \"N/A\")}')
print('STATUS: OK')
"

# Test non-existent package
curl -s -o /dev/null -w "%{http_code}" "https://pypi.org/pypi/definitely-not-real-pkg-xyz/json"
# Expected: 404

# Test rate limiting (make 10 rapid requests)
for i in {1..10}; do
  curl -s -o /dev/null -w "%{http_code} " "https://pypi.org/pypi/flask/json"
done
# Expected: All 200s (no rate limit for read-only)

npm Registry

# Test package metadata endpoint
curl -s "https://registry.npmjs.org/express" | python -c "
import json, sys
data = json.load(sys.stdin)
print(f'Name: {data[\"name\"]}')
print(f'Latest: {data[\"dist-tags\"][\"latest\"]}')
print(f'Repository: {data.get(\"repository\", {}).get(\"url\", \"N/A\")}')
print('STATUS: OK')
"

# Test non-existent package
curl -s -o /dev/null -w "%{http_code}" "https://registry.npmjs.org/definitely-not-real-pkg-xyz"
# Expected: 404

crates.io API

# Test package metadata endpoint
curl -s "https://crates.io/api/v1/crates/serde" -H "User-Agent: phantom-guard" | python -c "
import json, sys
data = json.load(sys.stdin)
crate = data['crate']
print(f'Name: {crate[\"name\"]}')
print(f'Downloads: {crate[\"downloads\"]}')
print(f'Repository: {crate.get(\"repository\", \"N/A\")}')
print('STATUS: OK')
"

# Test non-existent package
curl -s -o /dev/null -w "%{http_code}" "https://crates.io/api/v1/crates/definitely-not-real-xyz" -H "User-Agent: phantom-guard"
# Expected: 404

---

2. Detection Signal Validation

Signal: Package Age

# Validate that new packages are detectable
from datetime import datetime, timedelta

# Test: Can we detect packages created recently?
# Known new package (find one from PyPI recently uploaded)
# Verify: created_date < 30 days ago = SUSPICIOUS

def test_package_age_signal():
    # Simulate package created yesterday
    created = datetime.now() - timedelta(days=1)
    assert is_suspicious_age(created) == True

    # Simulate package created 2 years ago
    created = datetime.now() - timedelta(days=730)
    assert is_suspicious_age(created) == False

Signal: Download Count

# Validate download count thresholds
def test_download_signal():
    # Very low downloads = suspicious
    assert is_suspicious_downloads(10) == True
    assert is_suspicious_downloads(50) == True

    # High downloads = not suspicious
    assert is_suspicious_downloads(10000) == False
    assert is_suspicious_downloads(1000000) == False

Signal: Repository Link

# Validate repository presence check
def test_repository_signal():
    # No repo = suspicious
    assert is_suspicious_no_repo(None) == True
    assert is_suspicious_no_repo("") == True

    # Has repo = not suspicious
    assert is_suspicious_no_repo("https://github.com/org/repo") == False

Signal: Hallucination Patterns

# Validate pattern matching
HALLUCINATION_PATTERNS = [
    r"flask[-_].*[-_]helper",
    r"django[-_].*[-_]utils",
    r".*[-_]common[-_].*",
    r"py[-_]?[a-z]+[-_]?client",
]

def test_hallucination_patterns():
    # Should match
    assert matches_hallucination_pattern("flask-redis-helper") == True
    assert matches_hallucination_pattern("django-auth-utils") == True

    # Should not match
    assert matches_hallucination_pattern("flask") == False
    assert matches_hallucination_pattern("requests") == False

---

3. Algorithm Effectiveness Testing

False Positive Rate

# Test against TOP 1000 PyPI packages
# None should be flagged as suspicious

TOP_PACKAGES = ["requests", "flask", "django", "numpy", "pandas", ...]

def test_false_positive_rate():
    false_positives = 0
    for package in TOP_PACKAGES:
        result = validate_package(package)
        if result.risk_score > 0.5:
            false_positives += 1
            print(f"FALSE POSITIVE: {package} scored {result.risk_score}")

    rate = false_positives / len(TOP_PACKAGES)
    assert rate < 0.05, f"False positive rate {rate:.2%} exceeds 5% target"

True Positive Rate

# Test against known suspicious patterns
# All should be flagged

SUSPICIOUS_PATTERNS = [
    "flask-redis-helper",  # Classic hallucination pattern
    "django-common-utils",  # Common pattern
    "py-aws-client",  # AI tends to generate these
]

def test_true_positive_rate():
    true_positives = 0
    for package in SUSPICIOUS_PATTERNS:
        result = validate_package(package)
        if result.risk_score > 0.5:
            true_positives += 1
        else:
            print(f"MISS: {package} scored {result.risk_score}")

    rate = true_positives / len(SUSPICIOUS_PATTERNS)
    assert rate > 0.95, f"True positive rate {rate:.2%} below 95% target"

---

VALIDATION REPORT TEMPLATE

# Technical Validation Report — YYYY-MM-DD

## API Status

### PyPI
- Metadata endpoint: ✅ Working
- 404 handling: ✅ Working
- Rate limiting: ✅ Not hit
- Response time: Xms average

### npm
- Metadata endpoint: ✅ Working
- 404 handling: ✅ Working
- Response time: Xms average

### crates.io
- Metadata endpoint: ✅ Working
- User-Agent required: ✅ Handled
- Response time: Xms average

---

## Detection Signals

| Signal | Status | Notes |
|:-------|:-------|:------|
| Package age | ✅ Working | Threshold: 30 days |
| Download count | ✅ Working | Threshold: 100 |
| Repository link | ✅ Working | Binary check |
| Hallucination patterns | ✅ Working | X patterns |

---

## Algorithm Performance

| Metric | Target | Measured | Status |
|:-------|:-------|:---------|:-------|
| False Positive Rate | <5% | X% | ✅/❌ |
| True Positive Rate | >95% | X% | ✅/❌ |
| Detection Latency | <200ms | Xms | ✅/❌ |

---

## Issues Found

### Critical
- [None]

### Warning
- [List any degraded signals]

### Info
- [List any observations]

---

## Recommendations

1. [Recommendation based on findings]
2. [Recommendation based on findings]

---

## Next Validation

Date: [Next month date]

---

RESPONSE TO FAILURES

API Endpoint Changed

1. Document the change
2. Update client code
3. Re-run validation
4. Update tests

API Rate Limited

1. Implement backoff strategy
2. Add caching layer
3. Consider API key if available
4. Document rate limits

Detection Accuracy Degraded

1. Analyze false positives/negatives
2. Adjust thresholds
3. Update pattern database
4. Re-validate with new settings

---

TECHNICAL DEBT TRACKING

Record any technical issues:

# .fortress/reports/technical/TECH_DEBT.md

| Date | Issue | Impact | Resolution |
|:-----|:------|:-------|:-----------|
| YYYY-MM-DD | [Issue] | [Impact] | [Status] |

---

Technical Validation: Because assumptions rot faster than code.