@useveto/node

Runtime authorization for AI agents. The missing layer between identity and action.

What is Veto?

Veto intercepts every tool call your AI agent makes, evaluates it against your policies, and decides: allow, deny, or escalate. Sub-10ms. Default deny.

• Default deny — no matching policy = blocked
• MCP native — drop-in middleware for Model Context Protocol servers
• Full audit trail — every decision logged
• Edge-first — powered by Cloudflare Workers

Install

npm install @useveto/node

Quick Start

import { VetoClient } from "@useveto/node";

const veto = new VetoClient({ apiKey: process.env.VETO_API_KEY! });

// Check if an agent can perform an action
const result = await veto.authorize("support-bot", "send_email", {
  to: "[email protected]",
  subject: "Refund confirmation",
});

if (!result.allowed) {
  console.log(`Blocked: ${result.reason}`);
}

MCP Integration

Option 1: Guard wrapper (recommended)

Wraps your tool handler — denied actions never execute.

import { VetoClient, createVetoGuard } from "@useveto/node";
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { z } from "zod";

const veto = new VetoClient({ apiKey: process.env.VETO_API_KEY! });
const protect = createVetoGuard(veto, { agentId: "support-bot" });

const server = new McpServer({ name: "my-server", version: "1.0.0" });

server.tool(
  "send-email",
  { to: z.string(), subject: z.string(), body: z.string() },
  protect("send-email", async (params) => {
    await sendEmail(params);
    return { content: [{ type: "text", text: "Email sent!" }] };
  }),
);

When denied, returns { content: [{ type: 'text', text: 'Authorization denied: ...' }], isError: true }.

Option 2: Manual middleware

Call at the start of your handler — throws VetoError if denied.

import { VetoClient, vetoMiddleware } from "@useveto/node";

const veto = new VetoClient({ apiKey: process.env.VETO_API_KEY! });
const guard = vetoMiddleware(veto, { agentId: "support-bot" });

server.tool("send-email", schema, async (params) => {
  await guard("send-email", params); // throws VetoError if denied
  await sendEmail(params);
  return { content: [{ type: "text", text: "Sent!" }] };
});

API Reference

VetoClient

const veto = new VetoClient({
  apiKey: "veto_...", // required
  endpoint: "https://api.veto.tools", // optional (default)
  timeout: 5000, // optional, ms (default: 5000)
});

Authorization

veto.authorize(agentId, toolName, parameters?) → Promise<AuthorizationResult>

Agents

veto.createAgent({ name, description? }) → Promise<Agent>
veto.listAgents() → Promise<Agent[]>
veto.getAgent(agentId) → Promise<Agent>
veto.deleteAgent(agentId) → Promise<void>

Policies

veto.createPolicy({ agentId, name, rules, priority?, enabled? }) → Promise<Policy>
veto.listPolicies(agentId?) → Promise<Policy[]>
veto.getPolicy(policyId) → Promise<Policy>
veto.updatePolicy(policyId, { name?, rules?, priority?, enabled? }) → Promise<Policy>
veto.deletePolicy(policyId) → Promise<void>

Audit Logs

veto.queryAuditLog({ agentId?, action?, toolName?, result?, from?, to?, limit?, offset? }) → Promise<AuditLogEntry[]>

Error Handling

import { VetoError, UnauthorizedError, RateLimitError } from "@useveto/node";

try {
  await veto.authorize("agent", "tool");
} catch (error) {
  if (error instanceof RateLimitError) {
    console.log(`Retry after ${error.retryAfterMs}ms`);
  } else if (error instanceof UnauthorizedError) {
    console.log("Invalid API key");
  } else if (error instanceof VetoError) {
    console.log(`${error.code}: ${error.message}`);
  }
}

Fail-Closed by Default

If Veto is unreachable (network error, timeout), the SDK denies by default. You can override this:

const protect = createVetoGuard(veto, {
  agentId: "my-agent",
  onError: "allow", // fail-open (not recommended for production)
});

Links

• Website: veto.tools
• Dashboard: app.veto.tools
• API Docs: docs.veto.tools
• GitHub: github.com/useveto/node

License

MIT