Refund a paid invoice (full or partial). Endpoint: POST /invoices/:id/refund
AI agents use refund_invoice to commit financial operations through Mcp Ap2 — usually the final step of a payment, billing, or trading workflow. A call moves real money.
This tool directly moves money by refunding payments. Misuse allows an AI agent to initiate refunds on invoices without proper authorization, causing financial loss. This is a core financial operation with high blast radius and is the most severe risk category applicable.
From the tool's definition Tool name is 'refund_invoice' and description states it refunds a paid invoice (full or partial) via POST to /invoices/:id/refund. The server itself is explicitly for 'Agent-to-Agent Payment Protocol.'
Documented attack patterns abuse exactly the kind of access refund_invoice gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Mcp Ap2, and nothing reaches the server without passing your rules. This is the rule we recommend for refund_invoice:
{
"version": "1",
"default": "deny",
"tools": {
"refund_invoice": {
"deny_if": [
{
"conditions": [],
"on_deny": "Requires human approval."
}
]
}
}
} Any call to refund_invoice is blocked until a human approves it. The rest of the server keeps working.
Free to start. No card required.
Refund a paid invoice (full or partial). Endpoint: POST /invoices/:id/refund. It is categorised as a Financial tool in the Mcp Ap2 MCP Server, which means it involves financial transactions. Block by default and require explicit approval.
Register the Mcp Ap2 MCP server in PolicyLayer and add a rule for refund_invoice: 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 Mcp Ap2. Nothing to install.
refund_invoice is a Financial tool with critical risk. Critical-risk tools should be blocked by default and only enabled with explicit human approval.
Yes. Add a rate_limit block to the refund_invoice 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 refund_invoice. 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.
refund_invoice is provided by the Mcp Ap2 MCP server (@codespar/mcp-ap2). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Mcp Ap2, 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.
1300 Mcp Ap2 tools catalogued and risk-classified — across an index of 43,000+ MCP servers.