Google Workspace CLI (gws)

Google Workspace CLI (official Google release) for Drive, Gmail, Calendar, Sheets, Docs, Chat, Admin, and every Workspace API. Includes native MCP server mod...

This skill appears to be the Workspace CLI described, but proceed cautiously: 1) Verify provenance — confirm the npm package @googleworkspace/cli and the GitHub repo are the official googleworkspace organization and check release signatures or commit history before installing. 2) Credentials — the CLI needs OAuth or a service-account JSON (sensitive). The SKILL.md references GOOGLE_WORKSPACE_CLI_CREDENTIALS_FILE but the registry metadata does not declare it; make sure you only provide least-privilege credentials and store them in a secure place (secret manager or isolated host). 3) Limit exposure — if you run gws mcp, restrict services with -s and use --tool-mode compact to reduce number of tools the agent can call. Prefer running the MCP server on an isolated machine or container rather than on a shared agent host. 4) Principle of least privilege — create service accounts with only the scopes needed, avoid granting broad Gmail/Drive scopes unless necessary, and rotate keys regularly. 5) Ask the publisher/maintainer for clarification: update registry metadata to declare required env vars (e.g., GOOGLE_WORKSPACE_CLI_CREDENTIALS_FILE), provide an authoritative install spec and checksums, and confirm that the package is the official Google release. If you cannot confirm provenance or cannot enforce least-privilege credentials and environment isolation, treat this skill as high-risk and avoid installing it in production environments.

Type: OpenClaw Skill Name: skill-gws Version: 1.0.0 The skill provides an interface for the official Google Workspace CLI (@googleworkspace/cli), enabling AI agents to interact with services like Gmail, Drive, and Calendar via the Model Context Protocol (MCP). The installation and authentication processes (OAuth/Service Accounts) are transparently documented in SKILL.md and align with the tool's stated purpose. No evidence of malicious intent, data exfiltration to unauthorized endpoints, or prompt injection was found.