Budgets, payee allowlists, price-spike checks and human approval for agents paying x402 paywalls
Spend Guard is an MCP integration. Budgets, payee allowlists, price-spike checks and human approval for agents paying x402 paywalls.
The published npm package is @vulinh1704/spend-guard, version 0.2.0. Its declared MCP transport is stdio.
The package declares SPEND_GUARD_POLICY: Policy JSON file: networks, budgets, approval amount, spike ratio, known services and their payees. Without one, every payment waits for your approval.; SPEND_GUARD_RAILS: How to pay: mock (pays nothing), near, arc, gateway, or a comma list. arc and gateway need ARC_PRIVATE_KEY, near needs NEAR_ACCOUNT_ID and NEAR_PRIVATE_KEY.; SPEND_GUARD_LEDGER: Decisions file, read back on start so a restart keeps today's spend; servers on one file share one budget. Default ~/.spend-guard/ledger.jsonl; off keeps it in memory.; SPEND_GUARD_DASHBOARD_PORT: Port of the approval dashboard on 127.0.0.1. 0 picks a free one; off means no dashboard, and payments that need approval are declined.; SPEND_GUARD_APPROVAL_TIMEOUT: Seconds a payment waits for your answer before it is declined. Raise it together with your client's tool timeout.; SPEND_GUARD_MOCK_NETWORK: Network the mock rail pretends to pay on.; ARC_PRIVATE_KEY: Agent key for the arc and gateway rails. Testnet keys only.; NEAR_ACCOUNT_ID: Agent account for the near rail (testnet).; NEAR_PRIVATE_KEY: Agent key for the near rail. Testnet keys only..
This package has launch arguments in its registry record. Use the maintainer’s configuration for those arguments rather than launching it with only the package name.
Check the maintainer’s instructions for account access, authentication and the supported client configuration. Once connected, inspect the tools your client receives and try a small task that matches the server’s stated purpose. The registry description does not include a verified, complete tool list.
View the publisher’s registry entry · Project repository · Project website. This profile uses version 0.2.0 from the registry. Registration identifies the publisher namespace; it does not establish that this is a first-party integration for every service it connects.
Browse related other integrations MCP servers for tools that cover the same kind of work.
| Variable | Description | Required |
|---|---|---|
| SPEND_GUARD_POLICY | Policy JSON file: networks, budgets, approval amount, spike ratio, known services and their payees. Without one, every payment waits for your approval. | Check documentation |
| SPEND_GUARD_RAILS | How to pay: mock (pays nothing), near, arc, gateway, or a comma list. arc and gateway need ARC_PRIVATE_KEY, near needs NEAR_ACCOUNT_ID and NEAR_PRIVATE_KEY. | Check documentation |
| SPEND_GUARD_LEDGER | Decisions file, read back on start so a restart keeps today's spend; servers on one file share one budget. Default ~/.spend-guard/ledger.jsonl; off keeps it in memory. | Check documentation |
| SPEND_GUARD_DASHBOARD_PORT | Port of the approval dashboard on 127.0.0.1. 0 picks a free one; off means no dashboard, and payments that need approval are declined. | Check documentation |
| SPEND_GUARD_APPROVAL_TIMEOUT | Seconds a payment waits for your answer before it is declined. Raise it together with your client's tool timeout. | Check documentation |
| SPEND_GUARD_MOCK_NETWORK | Network the mock rail pretends to pay on. | Check documentation |
| ARC_PRIVATE_KEY | Agent key for the arc and gateway rails. Testnet keys only. | Check documentation |
| NEAR_ACCOUNT_ID | Agent account for the near rail (testnet). | Check documentation |
| NEAR_PRIVATE_KEY | Agent key for the near rail. Testnet keys only. | Check documentation |
Run an Actual Budget ledger from your AI agent, with phone approvals for every change on paired machines.
An Odoo module that serves a read-only MCP endpoint at /mcp, scoped to each user's own access rights.
Scans a Windows screen into a numbered element table so an agent clicks and types by id instead of guessing pixels.
Targeted .NET assembly inspection for AI agents on Windows: find types, decompile one method, trace callers.
Agent memory that keeps every correction with its history, plus saved working state so a crashed run can resume.
Lets an AI assistant upload files, images and text to Pixeldrain and download them back, with optional client-side encryption.