google-workspace

Google Workspace CLI

Use gws when the task is about Google Workspace resources or Google Workspace API automation.

Prefer gws over generic HTTP calls when possible because it already knows the Workspace API surface and returns structured JSON by default.

Rules

1. Start with read/list/get commands before mutating Workspace state.
2. Confirm IDs first: document IDs, spreadsheet IDs, file IDs, message IDs, calendar IDs, space IDs.
3. Do not run interactive auth flows from an agent tool call. If auth is missing, report that gws needs to be configured in plugin settings or via a manual terminal login.
4. Keep commands machine-readable. Prefer JSON output and parse it instead of scraping human text.
5. For large list operations, limit the scope first, then page or filter.

Common Commands

Check installation and health:

gws doctor

Inspect help for a resource or method:

gws help
gws drive help
gws drive files help

Google Docs:

gws docs get --document-id <DOC_ID>

Google Drive:

gws drive files list --params '{"pageSize":10}'
gws drive files get --file-id <FILE_ID>

Google Sheets:

gws sheets spreadsheets get --spreadsheet-id <SPREADSHEET_ID>

Gmail:

gws gmail users messages list --user-id me --params '{"maxResults":10}'
gws gmail users messages get --user-id me --message-id <MESSAGE_ID>

Google Calendar:

gws calendar events list --calendar-id primary --params '{"maxResults":10,"singleEvents":true}'

Google Chat:

gws chat spaces messages list --parent spaces/<SPACE_ID>

Tool Usage In SwarmClaw

When using the google_workspace tool:

• Put the gws command after the binary into args, for example: {"args":["drive","files","list"],"params":{"pageSize":5}}
• Use params for --params
• Use jsonInput for --json
• Use pageAll: true when you intentionally want all pages
• Use dryRun: true before risky mutations if you are unsure

Error Handling