Evaluate a JavaScript expression in the browser page context. Dangerous patterns (fetch, XHR, cookies, storage, eval, Function) are BLOCKED by default to mitigate prompt injection; pass allowDangerous:true to override when the caller has verified the expression is trusted.
AI agents invoke browser_evaluate to trigger actions in Wmux. 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.
Executing arbitrary JavaScript in a browser page context is a code execution primitive. Even with default safety filters, the tool runs JS in a live browser context and the filters can be bypassed with allowDangerous:true. Misuse could exfiltrate data, manipulate the DOM, steal credentials, or perform unauthorized actions on any loaded page — critical blast radius.
From the tool's definition 'Evaluate a JavaScript expression in the browser page context' — executes arbitrary JavaScript; dangerous patterns like fetch, XHR, cookies, storage, eval, Function are explicitly called out; allowDangerous:true override available
Attacks that exploit this kind of access
Evaluate a JavaScript expression in the browser page context. Dangerous patterns (fetch, XHR, cookies, storage, eval, Function) are BLOCKED by default to mitigate prompt injection; pass allowDangerous:true to override when the caller has verified the expression is trusted. It is categorised as a Execute tool in the Wmux MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Wmux MCP server in PolicyLayer and add a rule for browser_evaluate: 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 Wmux. Nothing to install.
browser_evaluate 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_evaluate 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_evaluate. 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_evaluate is provided by the Wmux MCP server (openwong2kim/wmux). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
browser_evaluate is one line of Wmux's registry record.
The record carries the whole server: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.
Teams ship this data inside their own products. See what a licence covers →