A new command-line tool released on GitHub combines Workspace’s extensible APIs into a single interface. It also shows how seriously the company is taking the agent AI moment.
It’s called the tool in its documentation, describing it as “one CLI for all of Google Workspace, built for humans and AI agents.” gws. It provides unified command-line access to Gmail, Google Drive, Calendar, Docs, Sheets, Slides, Chat and many other Workspace services.
But the more obvious details are hidden in the instructions: the documents include specific integration instructions for OpenClaw, an open-source AI agent that went viral in late January and has since become a Rorschach test for where agent AI is headed.
Google’s decision to name-check OpenClaw in official documents, even unofficial official documents, is not something companies do by accident.
Why is a command line tool important for AI agents?
💜 of EU technology
The latest rumblings from the EU tech scene, a story from our wise founder Boris and some questionable AI art. Free in your inbox every week. Register now!
First GWSAn AI agent that wanted to search a Gmail inbox, extract a file from Drive, and update a Calendar event would have to navigate three separate APIs, each with its own authentication flow, rate limits, and response formats. The process worked, but as PCWorld described it, it was “a royal pain.”
The new tool packs it into a single interface. Each operation produces structured JSON output in a format that AI agents can reliably parse without the ambiguity that can mar graphical interfaces. Authentication is done once via OAuth, then inherited by any agent that calls the tool.
The architecture has a particularly elegant feature: gws does not send a list of static commands. Instead, it reads Google’s own Discovery Service at runtime and builds the entire command surface dynamically. Whenever Google adds a new API endpoint, the tool picks it up automatically.
No versions to update, no outdated documentation to contend with. For agents designed to operate over long time horizons, this self-renewal quality is no small convenience; is a meaningful reliability guarantee.
The repository also includes more than 100 pre-built “agent skills” that cover common Workspace workflows: uploading files to Drive with automatic metadata, adding data to Sheets, scheduling Calendar events, forwarding Gmail attachments, and similar operations.
These are the discrete, composable building blocks that agent frameworks like OpenClaw are designed to chain together.
OpenClaw connection
OpenClaw’s story has moved quickly. The project was published in November 2025 by Austrian software developer Peter Steinberger under the name Clawdbot, which prompted a trademark complaint from Anthropic.
After a brief stint as Moltbot, it settled on OpenClaw in late January 2026. Within a few weeks, users created 1.5 million agents using the platform; The GitHub repository has garnered nearly 200,000 stars. OpenClaw’s core principle is simple enough to fit on a business card: AI that actually does everything.
On February 14, Sam Altman announced that Steinberger was joining OpenAI to lead the next generation of personal agents. OpenClaw will transition to an independent open source foundation that will be supported by OpenAI. “Omar is taking over the world,” Steinberger said in a farewell post. “My next mission is to create an agent that even my mother can use.”
With OpenClaw integration instructions in the docs three weeks after Steinberger joined OpenAI, Google’s Workspace CLI landing in the middle of this story is the kind of timing that doesn’t seem coincidental. Whether this is a deliberate competing answer, an accidental release, or simply a developer at Google posting something that’s already in progress is unconfirmed.
What is clear is that a major platform company has now built infrastructure specifically to make its applications more useful for the open source agent ecosystem that OpenAI has just acquired the architect for.
MCP and the bigger picture
Besides OpenClaw, gws also acts as a Model Context Protocol server. MCP is an open standard for how artificial intelligence agents communicate with external tools, originally developed by Anthropic and now adopted throughout the industry. Escape gws mcp Workspace exposes APIs as structured tools that can be called natively by any MCP-compatible client, Claude Desktop, VS Code with AI extensions, or Google’s own Gemini CLI.
This MCP support is significant because it means the tool is not just an OpenClaw utility. This is the infrastructure for all classes of AI agents that are bundled in MCP by default. Google is effectively making Workspace a first-class citizen in the emerging agent ecosystem, regardless of what model or framework is doing the work.
One important caveat: Google’s documentation clearly states this gws “not an officially supported Google product”. It is published as a developer sample, meaning there is no guarantee of production service-level stability, security, or ongoing maintenance. For individual developers and practitioners, this is a manageable risk.
For enterprises considering deploying AI agents against live Workspace data, this is a significant limitation, especially given ongoing concerns about OpenClaw’s security model, which the Cisco research team found vulnerable to data exfiltration and rapid injection through malicious third-party skills.
What does Google signal?
Addy Osmani, director of Google Cloud AI, defined his team’s focus as building infrastructure for agent systems that can generate command-line inputs and manage structured outputs for complex workflows. The Workspace CLI fits this vision directly.
A more extensive sample is worth reading. Microsoft has Copilot tasks. OpenAI now has the architect of OpenClaw. Google has its own Gemini agent stack, and now has a CLI that can be read by any agent that speaks JSON and MCP, the most widely used productivity suite.
Competition over where enterprise AI agents live and what data they can access is accelerating, and the battleground increasingly appears to be the infrastructure beneath the applications rather than the applications themselves.
for now, gws is a GitHub repository with warnings. But the 14,000 stars it collected before most journalists noticed it shows that developers who build agents for a living already understand what this means.
