create-connector

name	create-connector
description	Scaffold a new connector directory with base template, manifest, credentials.json, and requirements.txt
argument-hint	<connector-name>
disable-model-invocation	true

Create Connector: Scaffold a New Connector

Arguments

$ARGUMENTS contains the connector name (required) and optionally --etl.

Parse $ARGUMENTS:

• If --etl is present, this is an ETL connector
• Everything else is the connector name
• Default is data warehouse (no flag)

Examples: snowflake, bigquery, coalesce --etl, talend --etl.

Step 1: Run the scaffold script

For DW connectors (default):

python scripts/create_connector.py <connector-name>

This creates connectors/<name>/ with:

• connector.py — base template with all stubs
• manifest.json — unique connector type identifier
• credentials.json — empty credentials template
• requirements.txt — empty driver file
• Dockerfile.extra — system dependency instructions (empty by default)

For ETL connectors (--etl):

python scripts/create_connector.py <connector-name> --etl

This is interactive — it prompts for terminology mappings (group/job/task labels) and an optional icon URL. Before running the script, ask the user for both:

1. Terminology — propose a mapping based on the vendor's terms (e.g., for Coalesce: group=Environment, job=Job, task=Node; for Azure Data Factory: group=Data Factory, job=Pipeline, task=Activity) and let the user confirm or adjust.
2. Icon URL — ask if they want a custom icon for the integration (shown in the Monte Carlo UI). Must be a publicly reachable image URL (SVG/PNG). Verify it returns HTTP 200 before using. Skip if they don't want one — it can be added later as the icon_url key in manifest.json (requires rebuilding the agent image).

Then answer the script's prompts with those values (pipe via stdin if running non-interactively).

This creates etl_connectors/<name>/ with:

• connector.py — Connector class with fetch_metadata and fetch_run_details stubs
• manifest.json — connector identity with terminology mapping
• credentials.json — vendor API credential template (not database credentials)
• requirements.txt — vendor SDK dependencies

Step 2: Verify

Confirm the directory was created and list its contents:

DW:

ls -la connectors/<name>/

ETL:

ls -la etl_connectors/<name>/

Step 3: Report and suggest next step

For DW connectors:

Connector <name> scaffolded. Next step: run /setup-connection <name> to install the database driver and implement the connection methods.

For ETL connectors:

ETL connector <name> scaffolded. Next step: run /implement-etl-connector <name> to research the vendor API and implement the connector methods.