Tollgate
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Tollgatesend a $50 payout to alice@example.com for invoice 4471"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Tollgate
Tollgate is an open-source firewall for AI agents that move money through PayPal. Agents call Tollgate — over MCP or REST — instead of calling PayPal directly, and every money-moving request passes through a deterministic policy engine plus an AI guard that can only make verdicts stricter, never looser, before it ever reaches PayPal's sandbox.
Architecture
flowchart LR
Agent[AI Agent] -->|MCP / REST| Gateway[apps/gateway]
Gateway --> Auth[Authn + idempotency]
Auth --> Policy[Policy Engine<br/>packages/policy]
Policy --> Guard[AI Guard<br/>packages/guard-ai]
Guard -->|allow| PayPal[PayPal Sandbox<br/>packages/paypal]
Guard -->|hold| Approval[Approval Request + ntfy]
Guard -->|block| Reject[Reject + reasons]
Approval --> Human[Human approves on phone]
Human --> PayPal
Gateway --> Audit[(Hash-chained audit log<br/>packages/db)]
Gateway -->|SSE| Dashboard[apps/dashboard]Related MCP server: Parimit
Quickstart
TODO: docker compose up, pnpm demo, PayPal sandbox credentials.
Demo
TODO: 2-minute walkthrough script — legit payout, injected invoice blocked, held payout approved from phone, kill switch.
How AI is used
TODO: NL→policy compiler, injection scan, intent check — and why the AI can only escalate, never relax, a verdict.
PayPal APIs used
TODO: Payouts, Captures Refund, Invoicing v2, Orders v2 — sandbox only.
Security model
TODO: trust boundaries, fail-closed rules, hash-chained audit log — see docs/ARCHITECTURE.md.
License
MIT — see LICENSE.
This server cannot be deployed
Maintenance
Related MCP Connectors
Pre-spend firewall for AI agents. Approves, blocks, flags transactions against policy rules.
Payment infrastructure for AI agents: spending rules, approval flows, single-use virtual cards.
Runtime AI governance: decision gates, human approval, hash-chained audit, compliance mapping.
Human-in-the-loop review and approval for AI agents. Audit trail, approval policies, native MCP.
Related MCP Servers
- FlicenseNot gradedqualityCmaintenanceEnables AI agents to propose wallet payments while a local, human-authored policy decides whether each transaction is approved, requires human confirmation, or is refused, and records every decision in a signed, append-only ledger.-
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to propose structured payment intents, apply deterministic policy, require human approval, and record tamper-evident audit trails without connecting to live payment rails.MIT
- AlicenseNot gradedqualityBmaintenanceEnables store owners to reorder from suppliers by voice or chat and settle payments through PayPal, while a deterministic policy engine enforces spending limits, allow-listed suppliers, price-jump and duplicate checks so the LLM can only propose payments, never move money. Funds are authorized rather than charged, then captured only for goods verified on delivery via a 3-way match, with step-up human approval for unusual orders and a full audit ledger of every authorization, capture, void and refund.MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to carry out PayPal payments, invoice recordings, and refunds through a fail-closed firewall that enforces spending limits and trusted recipients, cross-checks requests against account records, and screens for prompt injection. Unusual or unverifiable requests are halted and escalated to a human for approval before any execution.MIT