Control a persistent browser. Actions: navigate (open URL), click (by CSS selector or x,y coords), type (text into selector or focused element), press (key like Enter/Tab), screenshot (capture current page), wait (wait ms for page to load), get_text (extract page text), close. Browser stays open ...
AI agents invoke browse_web to trigger actions in A-Modular-Kingdom. 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.
This tool executes real browser interactions—navigating URLs, clicking elements, typing text, pressing keys—against live external websites. The effects depend entirely on the arguments and could include submitting forms, triggering purchases, logging into accounts, or causing other external side effects. The persistent browser session compounds risk since state carries across calls.
From the tool's definition 'Control a persistent browser. Actions: navigate (open URL), click (by CSS selector or x,y coords), type (text into selector or focused element), press (key like Enter/Tab), screenshot (capture current page), wait (wait ms for page to load), get_text (extract…
Documented attack patterns abuse exactly the kind of access browse_web gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and A-Modular-Kingdom, and nothing reaches the server without passing your rules. This is the rule we recommend for browse_web:
{
"version": "1",
"default": "deny",
"tools": {
"browse_web": {
"limits": [
{
"counter": "browse_web_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
} browse_web 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.
Free to start. No card required.
Control a persistent browser. Actions: navigate (open URL), click (by CSS selector or x,y coords), type (text into selector or focused element), press (key like Enter/Tab), screenshot (capture current page), wait (wait ms for page to load), get_text (extract page text), close. Browser stays open between calls. It is categorised as a Execute tool in the A-Modular-Kingdom MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the A-Modular-Kingdom MCP server in PolicyLayer and add a rule for browse_web: 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 A-Modular-Kingdom. Nothing to install.
browse_web 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 browse_web 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 browse_web. 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.
browse_web is provided by the A-Modular-Kingdom MCP server (masihmoafi/a-modular-kingdom). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from A-Modular-Kingdom, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
14 A-Modular-Kingdom tools catalogued and risk-classified — across an index of 43,000+ MCP servers.