fabric-copy-joblisted
Install: claude install-skill wardawgmalvicious/agent-config
# Fabric Copy job (Data Factory)
Copy job is the guided, **no-pipeline** data-movement item in Fabric Data Factory: many sources → many destinations with bulk copy, incremental copy, and CDC replication. The **Copy job item is GA**; create via `+ New Item → Copy job`. Item type: **`CopyJob`**.
There is **no local file-based artifact** (unlike TMDL/PBIR) — you script Copy jobs via the REST API / MCP, or edit the JSON definition in-product. When to reach for Copy job vs alternatives: Copy job for guided source→destination movement; **Mirroring** for zero-config near-real-time DB replication; **pipeline Copy activity** when you need broader orchestration.
## Copy modes
| Mode | Behaviour | Status |
|---|---|---|
| **Full copy** | Every run copies all data source→destination (one-time or recurring snapshot). | GA |
| **Incremental — watermark** | First run = full load; later runs copy only rows past the watermark. | GA |
| **Incremental — CDC** | First run = full load; later runs replay inserts/updates/**deletes** from the source change feed. | **Preview** |
Copy job tracks the last-successful-run state itself. **A failed run doesn't advance state** — the next run resumes from the last success (no data loss, no manual watermark bookkeeping).
### Watermark vs CDC — decision rubric
| Consideration | CDC | Watermark |
|---|---|---|
| Source prerequisite | CDC enabled + connector supports it | A monotonically-increasing column |
| Detects inserts / updates | Yes / Yes | Yes /