Executes a natural language instruction in an active browser session. Use task for the main instruction. Returns action outcome, URL, and agent thinking. For complex goals, call multiple times, verifying each step.
AI agents invoke execute_instruction to trigger actions in Nova Act. 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 enables an AI agent to control a web browser and perform arbitrary actions (clicks, form submissions, navigation, etc.) based on natural language prompts. While the tool itself cannot directly delete data or move money, it can trigger such actions through browser automation.
From the tool's definition Tool 'execute_instruction' explicitly executes arbitrary natural language instructions in a browser session with outcomes dependent on the instruction content.
Documented attack patterns abuse exactly the kind of access execute_instruction gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Nova Act, and nothing reaches the server without passing your rules. This is the rule we recommend for execute_instruction:
{
"version": "1",
"default": "deny",
"tools": {
"execute_instruction": {
"limits": [
{
"counter": "execute_instruction_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}execute_instruction 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.
Executes a natural language instruction in an active browser session. Use task for the main instruction. Returns action outcome, URL, and agent thinking. For complex goals, call multiple times, verifying each step. It is categorised as a Execute tool in the Nova Act MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the Nova Act MCP server in PolicyLayer and add a rule for execute_instruction: 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 Nova Act. Nothing to install.
execute_instruction 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 execute_instruction 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 execute_instruction. 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.
execute_instruction is provided by the Nova Act MCP server (madtank/nova-act-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Nova Act, 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.
9 Nova Act tools catalogued and risk-classified — across an index of 43,000+ MCP servers.