Connectors

A connector is anything Dinobase can read from (and often write back to): a SaaS API, a database, a file path, an MCP server, or a custom REST endpoint. Each connector becomes a schema in DuckDB.

Category	Examples	Sync needed?	Storage
SaaS APIs	Stripe, HubSpot, GitHub	Yes	dlt syncs to parquet
Databases	PostgreSQL, MySQL, Snowflake	Yes	dlt syncs to parquet
Files	Parquet, CSV, S3, GCS	No	DuckDB reads files directly
MCP servers	Any stdio/SSE/HTTP MCP server	Yes	Tool output cached as JSON views
Custom REST	Any REST API via local YAML config	Yes/live	dlt fetches, cached as JSON views

The add command

dinobase add <type> [credentials] [options]

Credential resolution order

1. CLI flags — --api-key sk_live_...
2. Environment variables — STRIPE_SECRET_KEY
3. Interactive prompt — asks if neither is set

# These are equivalent:
dinobase add stripe --api-key sk_live_...

export STRIPE_SECRET_KEY=sk_live_...
dinobase add stripe

Custom naming

By default, the connector name matches the type. Use --name for multiple instances:

dinobase add stripe --api-key sk_live_... --name stripe_prod
dinobase add stripe --api-key sk_test_... --name stripe_test

Creates separate schemas: stripe_prod.* and stripe_test.*.

Sync intervals

Set per-connector intervals (used with dinobase sync --schedule):

dinobase add stripe --api-key sk_live_... --sync-interval 30m
dinobase add hubspot --api-key pat-... --sync-interval 1h

Supported formats: 30s, 5m, 1h, 6h, 1d.

SaaS API connectors

Powered by dlt verified sources and REST API connectors.

dinobase add stripe --api-key sk_live_...
dinobase add hubspot --api-key pat-na1-...
dinobase sync

After syncing, data is stored as parquet and queryable as stripe.*, hubspot.*, etc.

See SaaS APIs reference for all supported services.

Database connectors

Connect via connection string (SQLAlchemy-compatible):

dinobase add postgres --connection-string postgresql://user:pass@host:5432/db
dinobase add mysql --connection-string mysql://user:pass@host:3306/db
dinobase add snowflake --connection-string snowflake://user:pass@account/db/schema
dinobase add sqlite --path /path/to/database.db

See Databases reference for all supported databases.

File connectors

No sync needed. DuckDB reads files at query time through views.

# Local directory of parquet files
dinobase add parquet --path ./data/events/ --name analytics

# Single CSV file
dinobase add csv --path ./exports/report.csv --name report

# S3
dinobase add parquet --path s3://bucket/prefix/ --name warehouse

# GCS
dinobase add parquet --path gs://bucket/data/ --name warehouse

Path resolution

Input	Behavior
Directory (./data/)	Finds all matching files recursively
Single file (./data/events.parquet)	One table from the file
Glob (./data/*.parquet)	Matches by pattern
S3 URL (s3://bucket/prefix/)	DuckDB reads from S3 at query time
GCS URL (gs://bucket/prefix/)	DuckDB reads from GCS at query time

Each file becomes a table named after its filename: events.parquet becomes the events table.

See Files reference for more details.

Custom REST connectors

Connect any REST API endpoint by writing a local YAML config. Data is fetched via dlt (handles auth, pagination) and cached as JSON files that DuckDB queries via read_json_auto().

dinobase connector create posthog_flags \
  --url "https://app.posthog.com/api/" \
  --endpoint "projects/123/feature_flags/" \
  --data-selector results
dinobase add posthog_flags --api-key phx_xxx
dinobase query "SELECT name, active FROM posthog_flags.feature_flags"

See the full Custom REST Connectors reference for the YAML format, auth types, fetch modes, and connector management commands.

MCP server connectors

Connect any MCP server (stdio, SSE, or streamable HTTP) as a connector. Dinobase auto-discovers read-only tools and syncs their output as SQL tables. For writes or parameterized calls, use dinobase mcp call or the Python API.

dinobase connector create posthog_mcp \
  --transport stdio \
  --command "npx -y @posthog/mcp-server"
dinobase sync posthog_mcp
dinobase query "SELECT * FROM posthog_mcp.list_projects LIMIT 10"

See the full MCP Server Connectors reference for the YAML format, tool-selection rules, and direct tool-call examples (CLI + Python).

---

Viewing configured connectors

# List all available connector types
dinobase connectors

# See what's actually connected and loaded
dinobase status --pretty

Removing a connector

Edit ~/.dinobase/config.yaml directly to remove a connector entry. Data stays in DuckDB until overwritten.