Create Public Connector Documentation
Goal
Generate the public-facing documentation for the {{source_name}} connector, targeted at end users.
Output Contract
Produce a Markdown file strictly following the standard template at templates/community_connector_doc_template.md as src/databricks/labs/community_connector/sources/{{source_name}}/README.md.
Documentation Requirements
Please use the code implementation as the source of truth.
Use the source API documentation to cover anything missing.
Always include a section about how to configure the parameters needed to connect to the source system.
If the
connector_spec.yamldeclares aconnection.oauthblock, document the connection as OAuth-based: the user registers an OAuth app in the source and supplies itsclient_id+client_secret(not a token); the connection obtains the token automatically. Cover how to create the OAuth app in the source and which scopes to grant. Describe the user experience, not the internal flow name — do not mentionm2m/u2m/u2m_per_userin end-user docs (users don't configure those; the flow is fixed by the spec). Let the spec'sflowdecide only what you describe:- client-credentials: the connection authenticates machine-to-machine — no browser login, no redirect URI to register.
- interactive: creating the connection opens a browser for the user to sign in and authorize, so the OAuth-app setup section MUST document the Authorized redirect URIs to register, covering both ways a connection can be created:
- Databricks UI flow (from the "Add Data" page): the provider redirects back to Databricks, so the user must add
https://<your-workspace-url>/login/oauth/http.html(substituting their own workspace URL). Omitting this causes aredirect_uri_mismatcherror in the consent popup. community-connectorCLI flow: the CLI listens on a local loopback, so registering the loopback host (e.g.http://localhost, which covers thehttp://127.0.0.1:<port>/callbackthe CLI picks at runtime) is sufficient — no workspace URL is involved.
- Databricks UI flow (from the "Add Data" page): the provider redirects back to Databricks, so the user must add
See the Gmail connector README for the reference pattern. (Only surface a flow choice if the connector genuinely supports more than one and the user must pick.)
AVOID mentioning internal implementation terms such as function or argument names from the
LakeflowConnect.