seal

Verify Seal AI decision receipts: signed evidence anyone can check, no account needed.

AI & Machine Learning DevOps & Infrastructure

我该使用它吗

描述质量

94%

模式完整度

77%

命名质量

94%

投毒风险

100%

权限匹配度

100%

协议合规性

100%

发现（2）

LOWTool 'get_sample_receipt' description lacks action verb在 get_sample_receipt 中
LOWTool 'get_transparency_proof' description lacks action verb在 get_transparency_proof 中

基于对工具定义和协议合规性的自动分析。

上下文开销

~1,157token 数（工具定义）

~636 B典型响应大小

对注意力有中等影响（占 128k 上下文窗口的 0.90%）

这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高，可用于其他任务的注意力就越少。

安装

一键安装

将以下内容添加到你的 `claude_desktop_config.json` 文件中：

{
  "mcpServers": {
    "seal": {
      "url": "https://app.aqta.ai/api/mcp"
    }
  }
}

远程端点

https://app.aqta.ai/api/mcpstreamable-http

它能做什么

工具清单

Check that a record still says what was signed: the permitted or refused decision, unchanged, under a trusted or embedded key. Handles both formats: ATTESTATION-v1 (a model-call decision) and ACTION-v1 (an agent tool action). No Aqta service in the check.

输入模式

{
  "type": "object",
  "properties": {
    "receipt": {
      "type": [
        "object",
        "string"
      ],
      "description": "The record, as a JSON object or a JSON string."
    },
    "public_key": {
      "type": "string",
      "description": "Base64url Ed25519 key to trust. Omit to use the key embedded in the record, which proves internal consistency only, never provenance."
    },
    "profile": {
      "type": "string",
      "description": "Usually omit this. The record's own signed version tag selects the format, and ATTESTATION-v1 is the default. Pass 'ACTION-v1' only to force an agent-action record to be checked as one."
    }
  },
  "required": [
    "receipt"
  ]
}

Read a receipt field by field, with stated limits. Does not check the signature.

输入模式

{
  "type": "object",
  "properties": {
    "receipt": {
      "type": [
        "object",
        "string"
      ]
    }
  },
  "required": [
    "receipt"
  ]
}

Fetch the published Seal signing key.

输入模式

{
  "type": "object",
  "properties": {}
}

A genuine signed receipt from the sample workspace (demo key), for trying verification.

输入模式

{
  "type": "object",
  "properties": {
    "outcome": {
      "type": "string",
      "enum": [
        "ALLOWED",
        "BLOCKED",
        "SUPPRESSED"
      ],
      "description": "Pick a sample with this outcome. A BLOCKED receipt was signed without the call reaching the provider."
    }
  }
}

The published sample evidence pack: fixed review question, bound receipts, disclosure.

输入模式

{
  "type": "object",
  "properties": {
    "pack_id": {
      "type": "string",
      "description": "Pack id. Only the published \"sample\" pack is available without an account."
    }
  }
}

Verify a review pack receipt by receipt. Each verdict comes from the signed envelope itself, never from stored state, and results are per-receipt, never an average.

输入模式

{
  "type": "object",
  "properties": {
    "pack": {
      "type": [
        "object",
        "string"
      ],
      "description": "A pack object, or \"sample\" for the published sample pack."
    },
    "public_key": {
      "type": "string",
      "description": "Key to pin every receipt against. Omit to use each envelope's embedded key."
    }
  }
}

The receipt format: fields, outcomes, verification commands, and what a signature does not establish. Use before describing the format.

输入模式

{
  "type": "object",
  "properties": {}
}

List your workspace receipts filtered by outcome, policy, model, or date range. Requires your aqta_ key as Bearer.

输入模式

{
  "type": "object",
  "properties": {
    "outcome": {
      "type": "string",
      "enum": [
        "ALLOWED",
        "BLOCKED",
        "SUPPRESSED",
        "PASSED"
      ],
      "description": "Only receipts with this outcome."
    },
    "policy": {
      "type": "string",
      "description": "Only receipts where this policy ran, e.g. pii_scan."
    },
    "model": {
      "type": "string",
      "description": "Only receipts for this model."
    },
    "since": {
      "type": "string",
      "description": "ISO timestamp lower bound."
    },
    "until": {
      "type": "string",
      "description": "ISO timestamp upper bound."
    },
    "limit": {
      "type": "number",
      "description": "Max rows, default 25."
    }
  }
}

Inclusion proof plus signed tree head for one of your receipts. Requires your aqta_ key.

输入模式

{
  "type": "object",
  "properties": {
    "attestation_id": {
      "type": "string"
    }
  },
  "required": [
    "attestation_id"
  ]
}

Outcome, model and policy counts across your recent receipts. Requires your aqta_ key.

输入模式

{
  "type": "object",
  "properties": {
    "limit": {
      "type": "number",
      "description": "How many recent receipts to summarise, default 100."
    }
  }
}

对比