Connections Overview

Manage credentials for databases, APIs, and external services with encrypted storage.

Manage credentials for databases, APIs, and external services. Passwords are encrypted at rest.

What are connections

A connection stores the credentials and host information needed to access an external system. Instead of hardcoding database URIs or API keys in pipeline configs, you reference a connection by its ID:

{"conn_id": "prod-postgres"}

The engine resolves the connection ID to a full URI at runtime.

Connection types

Type	Label	Use case
`postgres`	PostgreSQL	Database source/sink nodes
`mysql`	MySQL	Database source/sink nodes
`sqlite`	SQLite	Database source/sink nodes
`http`	HTTP / REST API	API source/sink nodes
`sftp`	SFTP / SSH	File transfers
`s3`	Amazon S3	Cloud storage
`generic`	Generic	Any TCP service

Encryption

All passwords and the extra field are encrypted with AES-256-GCM before storage. The encryption key is stored in {db_path}.key (e.g., broked.db.key).

Passwords are never returned in API responses (always ********)
The extra field (for headers, tokens, etc.) is encrypted and only decrypted when actually used
The encryption key is generated automatically on first run

Protect your key file: Back up the .key file. If lost, encrypted connection passwords cannot be recovered.

Creating connections

Go to Connections in the sidebar and click New Connection. Fill in the type, host, port, credentials, and click Save.

API

curl -X POST http://localhost:8080/api/connections \
  -H "Content-Type: application/json" \
  -d '{
    "conn_id": "prod-postgres",
    "type": "postgres",
    "host": "db.example.com",
    "port": 5432,
    "schema": "mydb",
    "login": "etl_user",
    "password": "secret123"
  }'

Python SDK

Connections are managed via the UI or API, not the SDK. Reference them in pipelines by conn_id.

Testing connections

Test connectivity before using a connection in a pipeline:

curl -X POST http://localhost:8080/api/connections/prod-postgres/test

Response:

{"success": true, "message": "Connected successfully (pgx)", "driver": "pgx"}

Connection fields

Field	Type	Description
`conn_id`	string	Unique slug identifier (required)
`type`	string	Connection type (required)
`host`	string	Hostname or URL
`port`	int	Port number
`schema`	string	Database name
`login`	string	Username
`password`	string	Password (encrypted at rest)
`extra`	string	JSON string for additional config (encrypted at rest)

Using connections in pipelines

Reference by conn_id in any node that supports it:

{
  "type": "source_db",
  "config": {
    "conn_id": "prod-postgres",
    "query": "SELECT * FROM users"
  }
}

The engine builds the full connection URI from the stored fields.

Usage tracking

See which pipelines use a connection:

curl http://localhost:8080/api/connections/prod-postgres/used-by

This helps prevent accidentally deleting connections that are in use.

On this page