Connect an AI agent to Google Workspace with gws
Prepare the account, choose the appropriate access and decide how the first command should return its result.
To connect an AI agent to Google Workspace with gws, use gws-shared as the common reference for authentication and command options. It describes interactive OAuth, authorised service-account credentials, output formats and local destinations. A short read from one intended service is a useful first request when checking which work account the agent will use.
Think about the output as well as the input. A small table can remain in the conversation, while a binary response can be saved to a named path. Optional screening can send a response to a configured Google Cloud Model Armor template. Pick the route that fits the account's permissions and the sensitivity of the material.
What it does
- Set up a documented gws authentication method
- Choose the service and requested operation
- Select a table or structured result format
- Preview supported write operations using dry-run
- Provide a local destination for binary output
Install and use
- Install the documented gws tool and the shared skill. Connect an authorised Google user or approved service account, enabling only the APIs and scopes needed for the job.
- Ask “For our first lookup, show me which service and account will be used and return a short table.”
- Check the destination before requesting a write, export or optional Model Armor screen. Keep credentials out of the conversation.
FAQ
Source
Made by Google Workspace CLI contributors. Original skill: gws-shared, published on GitHub under the Apache-2.0 licence.
View on GitHub