Skip to content
Documentation
Supported Data Sources

Supported Data Sources

PlyDB abstracts the complexity of different storage formats into a single relational view. You can query across any combination of these sources using standard SQL.

For configuration details, see Configuring Data Sources.

SQL Databases

PostgreSQL

Full support for querying PostgreSQL databases. PlyDB connects over the network using standard credentials. Tables are referenced as catalog.schema.table (e.g. analytics.public.orders).

PostgreSQL COMMENT metadata is automatically extracted for semantic context, giving AI agents rich descriptions of your tables and columns.

MySQL

Full support for querying MySQL databases with the same network-based connection model as PostgreSQL.

SQLite

Query SQLite database files directly by path. SQLite uses main as its default schema, so tables are referenced as catalog.main.table.

DuckDB

Query DuckDB database files directly by path. Like SQLite, DuckDB uses main as its default schema.

File Formats

PlyDB can query flat files as if they were database tables. The file format is inferred from the extension, or can be set explicitly in the config.

CSV

Comma-separated (or custom-delimited) value files. Supports configurable delimiter and header row detection.

JSON

JSON files containing tabular data.

Parquet

Apache Parquet columnar storage files. Commonly used for large analytical datasets and data lake exports.

Excel (.xlsx)

Excel workbooks. You can specify which sheet tab to read via the sheet_name config field.

Object Storage

Amazon S3

Query files stored in S3 buckets. Supports all file formats listed above (CSV, JSON, Parquet, Excel). S3 URIs support glob patterns for matching multiple files:

s3://bucket/2026/*/sensor_*.parquet

Requires AWS credentials configured via the credentials section of the config file.

SaaS

Google Sheets

Query Google Sheets spreadsheets using SQL. Two authentication methods are supported:

  • Service account — for server-side, non-interactive use.
  • Browser OAuth — for interactive, ad-hoc querying with no credentials in the config.

When sheet_name is omitted from the config, the table name in your SQL query determines which tab is read, allowing you to query different tabs without changing configuration.

Planned

The following sources are on the roadmap but not yet supported:

  • Apache Iceberg — Open table format for large analytical datasets.
  • Delta Lake — Storage layer for data lakehouse architectures.
Agent IntegrationFAQ