PlyDB CLI skill

The plydb CLI can be used to query across heterogenous data sources.

Dependencies

The plydb binary must be available on the system.

If it is not, installation instructions can be found here

Instructions

Configure data sources

First, the data sources to make available to PlyDB must be configured in a config file as per the specification in references\config_schema.md.

Query with SQL

Once you have a data source config file, PlyDB can query across all of the configured data sources. Use fully qualified table names: catalog.schema.table.

plydb query \
  --config path/to/config/file/config.json \
  "SELECT * FROM customers.default.customers c
   JOIN orders.default.orders o
   ON c.id = o.customer_id"

Fetching semantic context of the data

To provide context to understand the domain and write correct SQL - PlyDB can build and provide semantic context from database COMMENT metadata alongside column types and foreign keys as structured YAML that follows the Open Semantic Interchange (OSI) specification.

plydb semantic-context --config path/to/config/file/config.json

Enriching auto-scanned context with overlays

When the database lacks comments or you need to add relationships and metrics not captured from source metadata, use --semantic-context-overlay to supply one or more OSI YAML files that are merged on top of the auto-scanned model:

plydb semantic-context \
  --config path/to/config/file/config.json \
  --semantic-context-overlay path/to/overlay.yaml

The flag is repeatable; overlays are applied in the order given:

plydb semantic-context \
  --config path/to/config/file/config.json \
  --semantic-context-overlay base_overlay.yaml \
  --semantic-context-overlay team_overlay.yaml

Overlay files must be valid Open Semantic Interchange (OSI) YAML.

Overlays can add descriptions to existing datasets and fields, define relationships between existing datasets, and add or update metrics. They cannot introduce new datasets or fields - only enrich what was already discovered by the auto-scanner.

Good opportunities to create or edit an overlay file are when encountering a new dataset or after a session of data analysis with the user. These are great opportunities to distill your learnings about the data's semantics and record them into an overlay file for future sessions. Ask the user first.

Embedding overlays in the config file

Overlays can also be specified in the config file under semantic_context.overlays instead of (or in addition to) the CLI flag:

{
  "databases": { ... },
  "semantic_context": {
    "overlays": [
      "path/to/base_overlay.yaml",
      "path/to/team_overlay.yaml"
    ]
  }
}

With overlays in the config, no extra flags are needed:

plydb semantic-context --config path/to/config.json

Config-file overlays are applied before any --semantic-context-overlay flags.

Troubleshooting

• gsheet data source with interactive OAuth