Database Quick Reference

# Load a seed in Python
python -c "from database.schemas.seeds import load_seed; print(load_seed('leagues/mltt.seed.json'))"
 
# List all seeds
find database/schemas/seeds -name "*.seed.json" | wc -l
 
# Add new seed
vim database/schemas/seeds/leagues/new-league.seed.json
git add database/schemas/seeds/leagues/new-league.seed.json

# Seed all examples
uv run python scripts/seed.examples.py
 
# Seed specific category
uv run python scripts/seed.examples.py --category triage
 
# Clear and reseed
uv run python scripts/seed.examples.py --clear --category triage
 
# Check seeded data
psql $DATABASE_URL -c "SELECT category, COUNT(*) FROM \"FewShotExample\" GROUP BY category;"

# Development: Load seed
from database.schemas.seeds import load_seed
league = load_seed("leagues/mltt.seed.json")
 
# Production: Query examples
from database.storage.examples import FewShotExamplesAPI
api = FewShotExamplesAPI()
examples = await api.get_examples_for_prompt(
    prompt_text="Premium basketball partnership",
    prompt_type="triage",
    max_examples=3
)
 
# Direct Prisma query
from prisma import Prisma
db = Prisma()
await db.connect()
examples = await db.fewshotexample.find_many(
    where={"category": "triage", "quality_score": {"gte": 0.8}},
    take=5
)

database/
├── schemas/
│   ├── seeds/              # Development test data (240+ files)
│   │   ├── leagues/        # 101 league definitions
│   │   ├── questionnaires/ # 53 questionnaire examples
│   │   ├── schemas/        # 105 schema seeds
│   │   ├── samples/        # 5 sample datasets
│   │   └── synthetic_emails/ # 34 test emails
│   │
│   └── examples/
│       └── few_shot/       # Future: Schema examples
│
└── storage/
    └── examples/           # Production few-shot KB
        ├── data/           # JSONL files (11 files, 349 lines)
        ├── api.py          # High-level API
        ├── retriever.py    # Core retrieval
        ├── matcher.py      # Similarity matching
        └── cache.py        # LRU cache

SELECT example_id, category, quality_score, usage_count
FROM "FewShotExample"
WHERE quality_score > 0.85
ORDER BY usage_count DESC
LIMIT 10;

SELECT 
    category,
    COUNT(*) as total,
    COUNT(CASE WHEN quality_score >= 0.8 THEN 1 END) as high_quality,
    AVG(quality_score) as avg_quality
FROM "FewShotExample"
GROUP BY category;

SELECT example_id, category, quality_score
FROM "FewShotExample"
WHERE usage_count = 0
  AND created_at < NOW() - INTERVAL '30 days'
ORDER BY quality_score DESC;

# Check file exists
ls database/schemas/seeds/leagues/mltt.seed.json
 
# Validate JSON
python -m json.tool database/schemas/seeds/leagues/mltt.seed.json

# Check DB connection
echo $DATABASE_URL
 
# Check seeded count
psql $DATABASE_URL -c "SELECT COUNT(*) FROM \"FewShotExample\";"
 
# Reseed if needed
uv run python scripts/seed.examples.py --clear

from database.storage.examples import ExampleCache
cache = ExampleCache()
cache.clear()  # Reset cache