# CarbonOps Dataset Schema

## 1) grid-intensity/
CSV/Parquet with columns:
- `timestamp` (UTC, ISO8601)
- `region` (string, snake_case)
- `intensity_gco2_per_kwh` (float, >=0)
- `source` (enum: `uk_national_grid`, `entsoe`, `custom_csv`)
- `source_url` (string, optional)
- `updated_at` (UTC)

Constraints:
- `timestamp` monotonically increasing within region
- one row per 30 min (GB) or configured resolution
- missing data → `NaN`, never `-1`

## 2) ci-samples/
- `job_id` (string), `repo` (string), `duration_s` (float), `artifact_mb` (float), `runner_type` (enum), `timestamp_start` (UTC).