Agent Reserve Protocol Authorization Firewall

Advisory policy preflight for AI-agent spend requests; never executes payments or accesses wallets.

我該用這個嗎

品質與安全性

A
說明品質
100%
結構描述完整度
83%
命名品質
100%
汙染風險
100%
權限相符程度
100%
協定合規性
100%

根據工具定義與協定合規性的自動化分析。

上下文成本

~213Token(工具定義)
~1.8 KB典型回應大小
極小的注意力影響(128k 上下文的 0.17%)

這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高,可用於其他工作的注意力就越少。

安裝

一鍵安裝

將以下內容加入你的 `claude_desktop_config.json` 檔案:

{
  "mcpServers": {
    "agent-transaction-firewall": {
      "url": "https://agent-reserve-protocol.super-myna-7255.chatgpt.site/api/mcp"
    }
  }
}

遠端端點

https://agent-reserve-protocol.super-myna-7255.chatgpt.site/api/mcpstreamable-http

它能做什麼

工具清單

工具(1)

🟢 唯讀🟡 寫入🔴 刪除⚪ 未知
🟢evaluate_agent_spend_request(mandateValid, vendorStatus, receiptStatus, amount, perRequestLimit, ...)

Return eligible, escalate, or block under an owner-defined mandate. Advisory preflight only; never executes payment or accesses a wallet.

輸入結構描述

{
  "type": "object",
  "properties": {
    "mandateValid": {
      "type": "boolean",
      "description": "Whether a current owner-approved mandate exists."
    },
    "vendorStatus": {
      "type": "string",
      "enum": [
        "trusted",
        "unverified",
        "blocked"
      ]
    },
    "receiptStatus": {
      "type": "string",
      "enum": [
        "pass",
        "missing",
        "fail"
      ]
    },
    "amount": {
      "type": "integer",
      "minimum": 1,
      "maximum": 1000000000
    },
    "perRequestLimit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 1000000000
    },
    "dailyRemaining": {
      "type": "integer",
      "minimum": 0,
      "maximum": 1000000000
    }
  },
  "required": [
    "mandateValid",
    "vendorStatus",
    "receiptStatus",
    "amount",
    "perRequestLimit",
    "dailyRemaining"
  ],
  "additionalProperties": false
}

社群

為此伺服器評分

證據

近期觀測

已驗證未記錄版本1 個工具
已驗證未記錄版本1 個工具