Invoke a tool on a downstream MCP server.
AI agents invoke gateway.invoke to trigger actions in PMCP. 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 acts as a gateway to execute any tool available on downstream servers. While the tool itself is neutral infrastructure, the ability to invoke arbitrary downstream tools means an AI agent with access to this could trigger Read, Write, Destructive, Financial, or other Execute operations depending on what downstream tools exist and what arguments are supplied.
From the tool's definition The tool description explicitly states it 'Invoke[s] a tool on a downstream MCP server.' Invocation of arbitrary tools on downstream servers is a form of code/command execution whose effects depend entirely on which tool is invoked and what arguments are…
Documented attack patterns abuse exactly the kind of access gateway.invoke gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and PMCP, and nothing reaches the server without passing your rules. This is the rule we recommend for gateway.invoke:
{
"version": "1",
"default": "deny",
"tools": {
"gateway.invoke": {
"limits": [
{
"counter": "gateway.invoke_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}gateway.invoke 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.
Invoke a tool on a downstream MCP server. It is categorised as a Execute tool in the PMCP MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the P MCP server in PolicyLayer and add a rule for gateway.invoke: 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 PMCP. Nothing to install.
gateway.invoke 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 gateway.invoke 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 gateway.invoke. 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.
gateway.invoke is provided by the P MCP server (viperjuice/pmcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from PMCP, 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.
26 PMCP tools catalogued and risk-classified — across an index of 43,000+ MCP servers.