Vantic MCP server

What it does

Vantic is a spending firewall for AI agents that enforces signed budgets and payment restrictions outside the model layer. It prevents prompt-injection attacks from causing unauthorized payments by checking transaction scope before execution and provides tamper-proof receipts for all actions.

• Restrict an autonomous agent to a weekly budget and per-transaction cap while giving it access to payment APIs.
• Wrap existing payment tools with spending guards that return errors to the model instead of executing out-of-scope charges.
• Verify a chain of signed receipts from agent transactions to produce an audit trail or dispute evidence.

Tools this server exposes

Tool	Does
mandate_authorize	Create and sign a spending mandate that defines budget, transaction limits, allowed actions, and counterparty allowlist for an agent.
mandate_verify_chain	Verify the cryptographic chain and integrity of receipts issued under a mandate.
mandate_verify_credential	Verify a W3C Verifiable Credential representing the owner-to-agent relationship.
mandate_resolve_did	Resolve a W3C DID to its document for identity verification.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.