Google has unveiled a new command-line tool, dubbed gws, designed to streamline access to its Workspace services – including Gmail, Drive, Calendar, Docs, and Sheets – for artificial intelligence agents. The release signals a growing recognition of the importance of agentic AI and a proactive step toward making its widely-used productivity suite more accessible to automated systems.
The tool, described in its documentation as “one CLI for all of Google Workspace, built for humans and AI agents,” provides a unified interface for interacting with the sprawling APIs that underpin Google’s online applications. But perhaps more revealing is the inclusion of dedicated integration guides for OpenClaw, the open-source AI agent that gained significant traction in early 2026, suggesting Google is paying close attention to the evolving landscape of AI-powered automation.
Before gws, AI agents attempting to manage tasks across multiple Workspace applications faced a fragmented experience, navigating separate APIs with differing authentication protocols and data formats. As PCWorld described it, this process was often “a royal pain.” The new tool aims to resolve this by providing a single point of access and standardized JSON output, making it easier for agents to parse and utilize information reliably.
A key architectural feature of gws is its dynamic nature. Unlike traditional command-line tools with static command lists, gws reads Google’s own Discovery Service at runtime, automatically updating its command surface whenever Google adds new API endpoints. This eliminates the need for manual updates and ensures the tool remains current with Google’s evolving ecosystem. According to the project’s GitHub repository, this means developers no longer need to wait for tool updates when Google launches new features.
The tool as well includes over 100 pre-built “agent skills” designed to cover common Workspace workflows, such as uploading files to Drive, appending data to Sheets, and scheduling Calendar events. These skills serve as building blocks for agent frameworks like OpenClaw, enabling the creation of complex, automated tasks.
The OpenClaw Connection
OpenClaw’s trajectory has been rapid. Initially launched in November 2025 by Austrian software developer Peter Steinberger as Clawdbot – a name that prompted a trademark complaint from Anthropic – the project was rebranded as Moltbot before settling on OpenClaw in January 2026. Within weeks, the platform had seen 1.5 million agents created by users, and its GitHub repository amassed nearly 200,000 stars. The project’s core premise – AI that “actually does things” – resonated with a growing community interested in practical applications of agentic AI.
In February 2026, OpenAI announced that Steinberger would be joining the company to lead the next generation of personal agents, with OpenClaw transitioning to an independent open-source foundation supported by OpenAI. Steinberger marked the occasion with a post stating, “The lobster is taking over the world,” and outlining his goal to build an agent accessible even to his mother.
The timing of the gws release, with its explicit OpenClaw integration instructions, three weeks after Steinberger’s move to OpenAI, has raised eyebrows. Even as Google has not confirmed whether this was a deliberate response to OpenAI’s acquisition of OpenClaw’s architect, the connection is undeniable. It demonstrates a clear effort by a major platform provider to build infrastructure that supports the open-source agent ecosystem.
MCP and the Broader AI Landscape
Beyond OpenClaw, gws also functions as a Model Context Protocol (MCP) server. MCP, an open standard for AI agent communication with external tools originally developed by Anthropic, is gaining traction across the industry. By running gws mcp, Workspace APIs are exposed as structured tools that can be natively accessed by any MCP-compatible client, including Claude Desktop, VS Code with AI extensions, and Google’s own Gemini CLI.
This MCP support positions gws as more than just an OpenClaw utility; it’s infrastructure for the broader class of AI agents converging on MCP as a standard. Google is effectively making Workspace a first-class citizen within the emerging agent ecosystem, regardless of the underlying model or framework.
Though, Google’s documentation explicitly states that gws is “not an officially supported Google product,” classifying it as a developer sample. This means there are no guarantees regarding stability, security, or ongoing maintenance at the level of a production service. While this may be acceptable for individual developers and experimentation, enterprises considering deploying AI agents against live Workspace data should be aware of the limitations. Concerns regarding OpenClaw’s security model, including vulnerabilities to data exfiltration and prompt injection identified by a Cisco research team, further underscore the need for caution.
What’s Next for Agentic AI and Google Workspace?
Addy Osmani, Director of Google Cloud AI, has emphasized his team’s focus on building infrastructure for agentic systems capable of generating command-line inputs and managing structured outputs. The Workspace CLI aligns directly with this vision. The competitive landscape is intensifying, with Microsoft’s Copilot Tasks, OpenAI’s acquisition of OpenClaw’s creator, and Google’s Gemini agent stack all vying for dominance. The battleground is shifting towards the infrastructure beneath applications, rather than the applications themselves.
Currently, gws remains a GitHub repository with a caveat. However, its rapid accumulation of over 14,000 stars suggests that developers building agents recognize its potential. As AI agents become more sophisticated and integrated into daily workflows, tools like gws will likely play an increasingly key role in bridging the gap between automation and productivity.
What are your thoughts on the rise of AI agents and their impact on productivity? Share your comments below.
