Skip to content
AdminKit

Tool allowlist

What is Tool allowlist?

A tool allowlist restricts an AI agent, or the gateway mediating its access, to a named, pre-approved set of tools it may call, rejecting any call outside that set even if the server exposes more. AdminKit's MCP/Agent Gateway enforces exactly this kind of allowlist, so a compromised or misconfigured server can't silently widen what an agent is allowed to do.

By AdminKit team, Netbase JSC · Updated

A tool allowlist is a security control that restricts an AI agent, or the gateway mediating its access, to a named, pre-approved set of tools it may call, rejecting any tool call outside that set even if the underlying server exposes more capability. This matters because an MCP or similar agent-tool server can expose far more capability than a given task needs, change its exposed tools over time, or be manipulated into offering a malicious tool under a trusted-sounding name; an allowlist means that even a compromised or misconfigured server can’t silently widen what an agent is able to do.

The MCP specification’s own security guidance says clients “SHOULD” prompt for user confirmation on sensitive tool calls and must treat tool metadata as untrusted unless it comes from a trusted server — an allowlist is the narrower, pre-decided version of that same control, fixed in advance rather than confirmed interactively on every call. OWASP’s GenAI Security Project lists this same pattern as a core mitigation for Excessive Agency (LLM06:2025), recommending teams “limit the extensions that LLM agents are allowed to call to only the minimum necessary.” A tool allowlist is a narrower, more auditable alternative to trusting an agent’s own judgment or a server’s self-reported tool list at call time. It’s a direct application of least privilege, which NIST defines as restricting the access privileges of users, or processes acting on their behalf, to the minimum necessary to accomplish assigned tasks. That same user-controlled review extends to a server’s exposed prompts, which are surfaced to the user rather than invoked directly by the agent, but whose underlying template can still carry the same kind of unreviewed instructions a malicious tool description would.

AdminKit groups tool allowlist under its AI agents & MCP cluster — browse the AdminKit glossary index for other related terms.

Frequently asked questions

What does a tool allowlist restrict?

An AI agent, or the gateway mediating its access, to a named, pre-approved set of tools it may call, rejecting any call outside that set.

Why does an allowlist matter even for a server that looks trustworthy?

A server can expose more capability than a given task needs, change its exposed tools over time, or be manipulated into offering a malicious tool under a trusted-sounding name.

How does an allowlist differ from the MCP specification's own confirmation guidance?

The spec says clients should prompt for user confirmation on sensitive tool calls; an allowlist is the narrower, pre-decided version of that same control, fixed in advance rather than confirmed interactively on every call.

What does OWASP recommend for mitigating Excessive Agency in LLM agents?

Limiting the tools or extensions an agent is allowed to call to only the minimum necessary, the same pattern a tool allowlist implements.