This record as markdown: /tools/cowork-bench/browser-close.md
What browser_close does on Cowork Bench
AI agents invoke browser_close to trigger actions in Cowork Bench. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.
Why browser_close is rated High
Closing a browser page is an action that affects an external process/session state. It is not purely destructive (no data deletion) nor read/write, but it executes an operation (terminating a browser page) that has real side effects such as losing unsaved state or interrupting workflows. Severity is medium because misuse could disrupt active browser sessions or cause loss of unsaved data.
From the tool's definition 'browser_close' and 'Close the page' indicate terminating a browser page/session, which is an external operation with side effects
Attacks that exploit this kind of access
The rule that runs browser_close safely
PolicyLayer is an MCP gateway: it sits between your AI agents and Cowork Bench, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For browser_close, this is the rule to start with:
browser_close stays usable, but rate-capped: a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
The button opens the PolicyLayer dashboard: create your workspace, connect Cowork Bench, apply this rule, and every browser_close call is checked against it from then on.
Questions about browser_close
Close the page. It is categorised as a Execute tool in the Cowork Bench MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Cowork Bench MCP server in PolicyLayer and add a rule for browser_close: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Cowork Bench. Nothing to install.
browser_close is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the browser_close rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for browser_close. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
browser_close is provided by the Cowork Bench MCP server (0717376/cowork_bench). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
More on Cowork Bench, and thousands of servers like it.
Across the catalogue