Execute a browser automation task using AI. EXECUTION MODE (default): - When skill_name is provided, hints are injected for efficient navigation. LEARNING MODE (learn=True): - Agent executes with API discovery instructions - On success, attempts to extract a reusable skill from the execution - If...
AI agents invoke run_browser_agent to trigger actions in Browser-Use MCP Server. 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 arbitrary browser automation tasks based on user-supplied natural language prompts. Browser automation can perform arbitrary clicks, form submissions, navigation, and interactions whose side effects depend entirely on the task argument.
From the tool's definition The tool description explicitly states 'Execute a browser automation task using AI' and can 'run', 'execute', and trigger external browser operations based on natural language task arguments.
Documented attack patterns abuse exactly the kind of access run_browser_agent gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Browser-Use MCP Server, and nothing reaches the server without passing your rules. This is the rule we recommend for run_browser_agent:
{
"version": "1",
"default": "deny",
"tools": {
"run_browser_agent": {
"limits": [
{
"counter": "run_browser_agent_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}run_browser_agent 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.
Execute a browser automation task using AI. EXECUTION MODE (default): - When skill_name is provided, hints are injected for efficient navigation. LEARNING MODE (learn=True): - Agent executes with API discovery instructions - On success, attempts to extract a reusable skill from the execution - If save_skill_as is provided, saves the learned skill Args: task: Natural language description of what to do in the browser max_steps: Maximum number of agent steps (default from settings) skill_name: Optional skill name to use for hints (execution mode) skill_params: Optional parameters for the skill (JSON string or dict) learn: Enable learning mode - agent focuses on API discovery save_skill_as: Name to save the learned skill (requires learn=True) Returns: Result of the browser automation task. It is categorised as a Execute tool in the Browser-Use MCP Server MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Browser-Use MCP Server MCP server in PolicyLayer and add a rule for run_browser_agent: 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 Browser-Use MCP Server. Nothing to install.
run_browser_agent 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 run_browser_agent 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 run_browser_agent. 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.
run_browser_agent is provided by the Browser-Use MCP Server MCP server (saik0s/mcp-browser-use). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 9 Browser-Use MCP Server tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
9 Browser-Use MCP Server tools catalogued and risk-classified — across an index of 42,500+ MCP servers.