--- name: cx-datasets description: > Discover the datasets a Coralogix team can query with the `cx datasets` CLI — the fully qualified `/` names DataPrime `source` commands need. Use when the user asks to "list datasets", "what datasets do we have", "which datasets can I query", "find datasets", "show dataspaces", "discover datasets", "what data does this team have", "available DataPrime sources", "filter datasets by pillar", "queryable sources", "dataset names for DataPrime", or wants to enumerate or filter the queryable datasets before running a DataPrime query. metadata: version: "0.1.0" --- # Datasets Skill Coralogix data lives in **datasets**, addressed as `/` — e.g. `default/logs`, `default/rum.events`, `system/labs.limitViolations`. Every DataPrime query starts with `source /`, so before querying an unfamiliar team (or a source you are not sure exists), list what is actually queryable instead of guessing names. ## CLI Commands | Command | Purpose | Key flags | |---|---|---| | `cx datasets list` | List every dataset queryable by the current profile's team | `--filter ` | | `cx datasets list -o json` | Same, as a JSON array of dataset rows | - | - Supports `-p ` (repeatable) for multi-profile fan-out; a `profile` key is added to each JSON row when more than one profile is targeted. - Read-only — no `--yes` confirmation needed. ## Dataset Row Fields | Field | Meaning | |---|---| | `dataspace` | The dataset's dataspace, e.g. `default` or `system` | | `dataset` | The dataset name inside its dataspace, e.g. `logs`, `rum.events` | | `pillar` | The pillar the data belongs to: `logs`, `spans`, `data`, ... | | `entityType` | The entity type behind the dataset, e.g. `logs`, `browserLogs/v2`, `jsonData` | | `version` | The latest dataset version | | `teamId` | The team that owns the dataset | Only read-enabled, latest-version datasets are returned. The `Source` column in text output is the copy-pasteable `source` argument. ## Workflow 1. **List the datasets** to see what exists: ```bash cx datasets list cx datasets list -o json | jq -r '.[] | .dataspace + "/" + .dataset' | sort ``` 2. **Narrow with a DPXL filter** when looking for a specific pillar or dataspace: ```bash cx datasets list --filter " \$d.pillar == 'logs'" cx datasets list --filter " \$d.dataspace == 'default' && \$d.pillar == 'data'" cx datasets list --filter " \$d.dataset.contains('rum')" -o json ``` 3. **Query the dataset** with DataPrime, using the full source name: ```bash cx dataprime query --source default/checkoutservice_dataset 'filter $d.severity == ERROR' cx dataprime query --source system/labs.limitViolations 'filter $m.timestamp > now()-1h' ``` `--source` is a `cx dataprime query` flag — `cx logs` and `cx spans` always query the legacy `logs`/`spans` datasets and take no source override. ## DPXL Filter Syntax `--filter` takes a DPXL expression **including its version tag** — it must start with ``. An expression missing the tag is rejected with a 400. In double-quoted shells (bash/zsh), escape the `$` as `\$d` so the shell does not expand it — the examples above are copy-paste safe. Available fields: `$d.dataspace`, `$d.dataset`, `$d.pillar`, `$d.entityType`, `$d.teamId`. Compare with `==` / `!=` and combine with `&&` / `||`. Substring matching uses method-call syntax — `$d.dataset.contains('rum')` (case-sensitive substring), not an infix `contains` operator. ## Key Principles - **Never guess dataset names** — a query against a wrong source fails or, worse, silently returns the wrong data. List first, then query. - **The source is `/`** — both parts are needed in DataPrime `source` commands; `default/logs` not just `logs`. - **Use `-o json` with `jq`** when you need just the source strings: `cx datasets list -o json | jq -r '.[] | .dataspace + "/" + .dataset'`. - **Filter server-side** with `--filter` rather than listing everything and grepping locally — same result, less output to parse. - **Multi-profile fan-out** with `-p ` (repeatable) lists datasets across environments in one call. ## Related Skills - **`cx-telemetry-querying`** — run the DataPrime queries against the datasets discovered here (logs, spans, metrics, RUM).