Anthropic integrates Claude browser extension with Cowork for cross-device session syncing

Anthropic has updated its Claude in Chrome browser extension to integrate with Claude Cowork, enabling seamless cross-device session synchronization. Users can now carry conversations, custom Skills, and external Connectors across desktop, web, and mobile apps without starting over. The extension can automate browser tasks like filling forms and clicking links, prompting Anthropic to introduce safety checks to mitigate prompt injection risks. The features are live for Max and Team plans, with Pro and Enterprise rollouts following.

Claude browser extension Cowork integration

• ▪Anthropic integrated its Claude in Chrome browser extension with Claude Cowork, allowing conversations, skills, and connectors to carry over between the browser and Claude's desktop, web, and mobile apps.
• ▪Claude Connectors link the AI to external tools and data sources like Google Drive, Gmail, and Slack through Model Context Protocol connections.
• ▪The Claude browser extension update, which arrived as part of a Chrome extension refresh on August 7, 2026, introduces support for Skills and Connectors that were previously confined to other parts of the Claude ecosystem.
• ▪Claude Skills are markdown-based instruction files that direct Claude's behavior in specific contexts, such as drafting legal contracts or enforcing coding styles.

Cross-device session synchronization

• ▪Claude sessions initiated in Cowork or through desktop tools like Claude Code maintain continuity when accessed through the Chrome browser extension, allowing users to switch devices without starting over.
• ▪Claude Cowork allows users to configure plugins for Skills, Connectors, and sub-agents on a per-conversation basis through desktop settings.

Browser automation capabilities

• ▪Anthropic expanded Claude's web interaction capabilities by adding a separate in-app browser for Claude Code that allows it to read documentation directly.
• ▪Claude in Chrome can view active webpages and perform actions such as clicking links, typing text, filling out forms, and navigating between pages using existing user logins.

AI agent security risks

• ▪Claude is programmed to request explicit user permission before performing high-risk actions such as making purchases or sharing personal information.
• ▪To mitigate browser automation risks, Anthropic implemented a check that reviews consequential actions, like submitting forms or sending messages, against the user's original request before execution.
• ▪Letting AI agents automate browser actions introduces security risks, particularly prompt injection attacks where hidden instructions in a webpage can redirect the agent without user knowledge.

Tier-based feature rollout

• ▪The Claude Cowork synchronization features are available immediately for Max and Team plans, with Pro access rolling out in the coming weeks.
• ▪Enterprise administrators can enable the new Claude browser features selectively by domain, while free-tier users will not receive access to these capabilities.