Google has simplified the use of Gmail and Drive for AI agents.

      A newly released command-line tool on GitHub streamlines Workspace’s extensive APIs into a singular interface. This move demonstrates the company's commitment to the rising trend of agentic AI. Named gws, this tool is described in its documentation as “one CLI for all of Google Workspace, built for humans and AI agents.” It offers unified command-line access to various Workspace services, including Gmail, Google Drive, Calendar, Docs, Sheets, Slides, and Chat.

      A noteworthy aspect is found in the setup instructions; the documentation features a specific integration guide for OpenClaw, the open-source AI agent that gained popularity in late January and has become somewhat of a litmus test for the direction of agentic AI. Google’s decision to mention OpenClaw in its official documentation, even if unofficial, is deliberate and unlikely to be a coincidence.

      The significance of a command-line tool for AI agents cannot be overstated. Prior to GWS, an AI agent seeking to search a Gmail inbox, extract a file from Drive, and update a Calendar event had to interact with three different APIs, each presenting unique authentication processes, rate limits, and response formats. Although functional, PCWorld referred to it as “a royal pain.”

      The new tool simplifies this by offering a single interface, with each operation resulting in structured JSON output that AI agents can reliably interpret, eliminating the confusion that often arises from graphical interfaces. Authentication occurs once through OAuth and is then inherited by any agent using the tool.

      One particularly impressive feature of the architecture is that gws does not come with a fixed list of commands. Instead, it dynamically reads Google’s Discovery Service in real-time, constructing its entire command surface on the fly. This means when Google adds a new API endpoint, the tool automatically incorporates it. There’s no need for version updates or outdated documentation; for agents intended to function over extended periods, this self-updating capability offers substantial reliability.

      The repository also contains over 100 pre-built “agent skills” that address common Workspace tasks, such as uploading files to Drive with automatic metadata, adding data to Sheets, scheduling Calendar events, forwarding Gmail attachments, and various other similar operations. These skills serve as individual, modular building blocks for agent frameworks like OpenClaw to combine.

      OpenClaw’s narrative has progressed rapidly. Initially published in November 2025 by Austrian software developer Peter Steinberger under the name Clawdbot, it faced a trademark issue from Anthropic. After briefly being named Moltbot, it rebranded to OpenClaw in late January 2026. Within weeks, users had created 1.5 million agents on the platform, and the GitHub repository received nearly 200,000 stars. The core concept of OpenClaw is straightforward: AI that takes action.

      On February 14, Sam Altman announced Steinberger’s transition to OpenAI to lead the development of the next generation of personal agents. OpenClaw would become an independent open-source foundation, supported by OpenAI. In his farewell post, Steinberger noted, “The lobster is taking over the world. My next mission is to build an agent that even my mum can use.”

      The timing of Google’s Workspace CLI launch coinciding with OpenClaw’s integration instructions—just three weeks after Steinberger joined OpenAI—appears purposeful. It remains unconfirmed whether this reflects a strategic competitive response, a coincidental release, or simply a continuation of ongoing development at Google. What is unmistakable is that a significant platform company is creating infrastructure aimed at enhancing its applications' compatibility with the open-source agent ecosystem that OpenAI has acquired.

      In addition to OpenClaw, gws serves as a Model Context Protocol (MCP) server, which is an open standard facilitating communication between AI agents and external tools, initially developed by Anthropic and now widely adopted in the industry. Running gws mcp makes Workspace APIs available as structured tools that any MCP-compatible client, such as Claude Desktop, VS Code with AI extensions, or Google’s own Gemini CLI, can utilize natively.

      The MCP integration is crucial because it indicates that the tool is not solely a utility for OpenClaw; it acts as infrastructure for the entire category of AI agents converging on MCP as a standard. Essentially, Google is positioning Workspace as a significant player within the emerging agent ecosystem, regardless of which model or framework is performing the tasks.

      However, a notable caveat is that Google’s documentation states gws is “not an officially supported Google product.” It is released as a developer sample, which means there are no guarantees regarding stability, security, or ongoing maintenance akin to a production service. For individual developers and experimenters, this represents a manageable risk. For enterprises contemplating the use of AI agents on live Workspace data, it is a considerable limitation, especially given ongoing concerns about OpenClaw’s security model, which a Cisco research team identified as vulnerable to data exfiltration and prompt injection through malicious third-party skills.

      Addy