dts

# DTS Query Workflow

This skill provides a structured workflow for executing and interpreting database queries with full LLM observability. All reasoning, queries, and interpretations are tracked via the DTS observer system for later review and human annotation.

## Quick Start

When exploring data or answering questions about dbt models:

1. **Think first** - Record your reasoning with `observer_context`
2. **Query with intent** - Execute queries with `db_exec` including why you're running them
3. **ALWAYS interpret** - Record what you learned with `db_interpret` **(REQUIRED)**
4. **Check for feedback** - Look for human annotations with `observer_get_annotations`

## Environment Context

The `dts-prime` command (auto-runs at session start) establishes schema context from dbt configuration. This context is critical for writing correct queries.

### Available Variables

After `dts-prime` runs, you have access to:

| Variable | Description | Example |
|----------|-------------|---------|
| `dev_schema` | Your development database.schema | `ANALYTICS_DEV.DEV_ADOVEN` |
| `prod_schemas` | Production database.schema(s) | `ANALYTICS.MARTS`, `ANALYTICS.FINANCE` |

### Query Patterns by Environment

**Dev Only** (testing your changes):
```sql
SELECT * FROM ANALYTICS_DEV.DEV_ADOVEN.orders LIMIT 100
```

**Prod Only** (production analysis):
```sql
SELECT * FROM ANALYTICS.MARTS.orders LIMIT 100
```

**Cross-Environment** (comparing dev to prod):
```sql
SELECT 'dev' as env, COUNT(*) as row_count
FROM ANALYTICS_DEV.DEV_ADOVEN.orders
UNION ALL
SELECT 'prod' as env, COUNT(*) as row_count
FROM ANALYTICS.MARTS.orders
```

### Always Use Fully-Qualified Names

**Do this:**
```sql
SELECT * FROM ANALYTICS.MARTS.orders
```

**Not this:**
```sql
SELECT * FROM orders  -- Relies on session context, will fail
```

See [references/ENVIRONMENTS.md](references/ENVIRONMENTS.md) for detailed patterns.

## Issue Context

When working on a tracked issue (GitHub, Linear, Jira, beads, etc.), include