The client_info constructor argument is the standard argument for changing user agent headers in GAPIC clients. Adding a user_agent prefix identifying an app or tool is required for many partners who use Google APIs.

Handwritten clients should provide a similar client_info argument to the client constructor. See how this was done in BigQuery at #7806.

Clients that need client_info:

• BigQuery (Add client_info to BigQuery constructor for user-amenable user agent headers #7806)
• Bigtable (Bigtable: add 'client_info' support to client. #7876)
• Datastore
• DNS (DNS: Add 'client_info' support to client / connection. #7869)
• Error Reporting (Error reporting: add 'client_info' support to Client. #7903)
• Firestore (and Firestore Beta) (Firestore: add client_info support to V1 client. #7877, V1 only, see below)
• Logging (Logging: add client_info support to client / connection. #7874, Logging: plumb 'Client._client_info' through to GAPIC API wrapper. #7901)
• Pub/Sub (Publisher, Subscriber) (done via **kwargs)
• Resource Manager (Resource Manager: Add 'client_info' support to client / connection. #7870)
• Runtime Config (Runtime Config: add client_info support to client / connection. #7871)
• Spanner - Note: Spanner has a user_agent argument already, but it's currently a no-op. Should mark that as deprecated. (Spanner: add client_info support to client. #7878)
• Storage (Storage: add client_info support to client / connection. #7872)
• Trace and Trace V1 (Trace: add client_info support to clients. #7899)
• Translate (Translate: add client_info support to client / connection. #7873, no GRPC version needed, because v3beta1 is pure-autogen)

I've heard requests for this feature in the Storage client, so I'd prioritize that one above the others, but we do need this feature across all clients.