Evidence row 221023

Observed 26 Oct 2026, 15:04 UTC, by probe.mcp/1.

Kind: mcp.tools
About: endpoint 476ddfbf-6081-45e9-bdce-67543d5f270f
Recorded: 26 Oct 2026, 15:04 UTC
Previous row: None: the first of its kind about this subject
Body SHA-256: 1f5fc3016e66cc58c861c69c9bec4de346f8182773e50fac121ee8f2e0844699
Leaf: 0x027730b9421d3459856b3b0bf68ecc5aa8b576bf31469bc6dd9d652d6e0bbcc6

What was observed

{
  "count": 5,
  "tools": [
    {
      "description": "Computer-Use Agent 动作风险分级。提交动作轨迹 → 逐步返回 L0(放行)/L1(放行+记录)/L2(需确认)/L3(阻断)/takeover(凭据场景挂起等用户亲手输入) 裁决、matched_rules、outcome，及 L0-L3 汇总计数；响应含 ruleset_version（规则库版本）与 disclaimer（本结果仅为分级参考，未执行实际拦截）。覆盖：敏感路径/破坏性命令/不可逆元素语义、域名分级（未知域导航 L2、黑名单域与敏感粘贴到未知域 L3、支付金额超阈值 L3）、凭据词表与支付/登录域 takeover、task_scope 任务意图偏离（越界升一级，只能收窄）、多智能体委托链字段（每条动作可选 chain_id/agent_id/parent_agent_id/delegation_depth，给了则服务端追加链级跨跳污点分析：跨 agent 传播时聚合级别抬到 L3 并置 risk_diluted，返回在 chain_trace 字段；未给则零影响）。纯规则判定，不调 LLM、不耗额度。用于在 agent 执行前/后做安全自检。",
      "inputSchema": {
        "$schema": "http://json-schema.org/draft-07/schema#",
        "properties": {
          "actions": {
            "description": "待分级的动作序列（1~500 条）",
            "items": {
              "description": "单条 CUA 动作",
              "properties": {
                "action": {
                  "description": "动作名（click/press/hotkey/write/move…），省略则按坐标推断",
                  "type": "string"
                },
                "action_input": {
                  "description": "ReAct 形态输入别名 → text",
                  "type": "string"
                },
                "agent_id": {
                  "description": "P1-9 本动作所属 Agent 名（如 research），配合 registered_agents 做注册校验",
                  "type": "string"
                },
                "app": {
                  "description": "target_app 别名",
                  "type": "string"
                },
                "chain_id": {
                  "description": "P1-9 多智能体委托链 id（uuid hex）；给了则审计按 chain_id 跨会话聚合，可在 /cua/audit 看链路视图",
                  "type": "string"
                },
                "coordinate": {
                  "additionalItems": false,
                  "description": "[x, y] 二元组（与 x/y 二选一）",
                  "items": [
                    {
                      "maximum": 9007199254740991,
                      "minimum": -9007199254740991,
                      "type": "integer"
                    },
                    {
                      "maximum": 9007199254740991,
                      "minimum": -9007199254740991,
                      "type": "integer"
                    }
                  ],
                  "maxItems": 2,
                  "minItems": 2,
                  "type": "array"
                },
                "delegation_depth": {
                  "description": "P1-9 委托深度（根 Agent=0）；超 delegation_max_depth 升 L2",
                  "maximum": 9007199254740991,
                  "minimum": 0,
                  "type": "integer"
                },
                "element": {
                  "description": "target_element 别名",
                  "type": "string"
                },
                "parent_agent_id": {
                  "description": "P1-9 委托方 Agent 名；跨 agent 边界的传播链由此配对",
                  "type": "string"
                },
                "path": {
                  "description": "target_path 别名",
                  "type": "string"
                },
                "target_app": {
                  "description": "目标应用",
                  "type": "string"
                },
                "target_element": {
                  "description": "目标元素标签",
                  "type": "string"
                },
                "target_path": {
                  "description": "文件操作路径 / 终端命令路径",
                  "type": "string"
                },
                "target_url": {
                  "description": "浏览器动作的目标 URL（域名分级 / 未知域导航 / 粘贴到未知域规则用）",
                  "type": "string"
                },
                "task_scope": {
                  "description": "P1-5 任务允许范围声明（服务端按动作读取；越界升一级，只能收窄）",
                  "properties": {
                    "allowed_apps": {
                      "description": "允许的前台应用白名单",
                      "items": {
                        "type": "string"
                      },
                      "type": "array"
                    },
                    "allowed_domains": {
                      "description": "允许访问的域名白名单（后缀匹配含子域）",
                      "items": {
                        "type": "string"
                      },
                      "type": "array"
                    },
                    "forbidden_elements": {
                      "description": "禁止出现的元素标签（最高优先级）",
                      "items": {
                        "type": "string"
                      },
                      "type": "array"
                    },
                    "task": {
                      "description": "声明的任务意图（如「整理本周会议纪要」）",
                      "minLength": 1,
                      "type": "string"
                    }
                  },
                  "required": [
                    "task"
                  ],
                  "type": "object"
                },
                "text": {
                  "description": "type/key/paste 的内容",
                  "type": "string"
                },
                "thought": {
                  "description": "agent 推理文本（审计留痕）",
                  "type": "string"
                },
                "url": {
                  "description": "target_url 别名",
                  "type": "string"
                },
                "x": {
                  "description": "坐标式点击 X",
                  "type": "number"
                },
                "y": {
                  "description": "坐标式点击 Y",
                  "type": "number"
                }
              },
              "type": "object"
            },
            "maxItems": 500,
            "minItems": 1,
            "type": "array"
          },
          "session_id": {
            "description": "给了则把分级结果写入审计表，可在 /cua/audit 回放",
            "maxLength": 64,
            "minLength": 1,
            "type": "string"
          },
          "task_scope": {
            "description": "任务意图对齐（收窄机制）：动作越出 scope 时升一级处置（L0→L1、L1→L2、L2 保持但追加原因、L3 不变）。只能升级不能降级；不传则零影响。",
            "properties": {
              "allowed_apps": {
                "description": "允许的前台应用白名单",
                "items": {
                  "type": "string"
                },
                "type": "array"
              },
              "allowed_domains": {
                "description": "允许访问的域名白名单（后缀匹配含子域）",
                "items": {
                  "type": "string"
                },
                "type": "array"
              },
              "forbidden_elements": {
                "description": "禁止出现的元素标签（最高优先级）",
                "items": {
                  "type": "string"
                },
                "type": "array"
              },
              "task": {
                "description": "声明的任务意图（如「整理本周会议纪要」）",
                "minLength": 1,
                "type": "string"
              }
            },
            "required": [
              "task"
            ],
            "type": "object"
          }
        },
        "required": [
          "actions"
        ],
        "type": "object"
      },
      "name": "check_cua_actions"
    },
    {
      "description": "CUA 代码审计：静态分析 Agent 源码，检测危险导入（subprocess/os.system 等）、权限边界（eval/exec）、注入面（字符串拼接 shell）、危险默认值、沙箱缺失。纯规则扫描，不调 LLM、不耗额度。支持单文件或多文件扫描。",
      "inputSchema": {
        "$schema": "http://json-schema.org/draft-07/schema#",
        "properties": {
          "code": {
            "description": "单文件模式：要扫描的源代码文本（≤50000 字符），与 files 二选一",
            "maxLength": 50000,
            "type": "string"
          },
          "filename": {
            "description": "单文件模式：文件名，用于报告中标识（默认 input.py）",
            "type": "string"
          },
          "files": {
            "description": "多文件模式：[{path, content, language?}] 数组，与 code 二选一",
            "items": {
              "properties": {
                "content": {
                  "maxLength": 50000,
                  "type": "string"
                },
                "language": {
                  "enum": [
                    "auto",
                    "python",
                    "javascript",
                    "typescript"
                  ],
                  "type": "string"
                },
                "path": {
                  "minLength": 1,
                  "type": "string"
                }
              },
              "required": [
                "path",
                "content"
              ],
              "type": "object"
            },
            "type": "array"
          },
          "language": {
            "description": "语言提示，auto 时按 filename 扩展名推断（默认 auto）",
            "enum": [
              "auto",
              "python",
              "javascript",
              "typescript"
            ],
            "type": "string"
          }
        },
        "type": "object"
      },
      "name": "check_cua_code_audit"
    },
    {
      "description": "AI 安全网关检测。对 prompt + output 跑 OWASP LLM Top 10 三层检测：40+ 特征规则引擎（注入/越狱/有害内容/敏感信息泄露/欺诈）+ 可选 LLM 深度分析 + 可选幻觉检测。返回 passed、risk_score、risk_level、threats[]（category/severity/match_context/recommendation）、quota。fast=true 走纯规则 <10ms 模式。耗 detect 额度（fast 亦计入）。",
      "inputSchema": {
        "$schema": "http://json-schema.org/draft-07/schema#",
        "properties": {
          "check_hallucination": {
            "description": "是否同时检测幻觉，默认 false",
            "type": "boolean"
          },
          "fast": {
            "description": "true 走纯规则快速模式（<10ms，无 LLM）→ /guard/check-fast；默认 false 走完整 /guard/check",
            "type": "boolean"
          },
          "output": {
            "description": "AI 响应（≥1 字，≤50000 字）",
            "maxLength": 50000,
            "minLength": 1,
            "type": "string"
          },
          "prompt": {
            "description": "用户原始输入（默认空，≤50000 字）",
            "maxLength": 50000,
            "type": "string"
          }
        },
        "required": [
          "output"
        ],
        "type": "object"
      },
      "name": "check_safety"
    },
    {
      "description": "Agent 输出与轨迹自检。对最终回答跑逐声明幻觉核验，并对执行轨迹做六维结构化评估（事实性/来源/指令合规/工具声明一致/任务完成/反思）。返回 main.claims、steps[].checks、quota。耗 detect 额度。",
      "inputSchema": {
        "$schema": "http://json-schema.org/draft-07/schema#",
        "properties": {
          "agent_output": {
            "description": "agent 最终输出文本（≥1 字，≤50000 字）",
            "maxLength": 50000,
            "minLength": 1,
            "type": "string"
          },
          "domain": {
            "description": "领域模式：general|medical|legal|finance|education|government",
            "enum": [
              "general",
              "medical",
              "legal",
              "finance",
              "education",
              "government"
            ],
            "type": "string"
          },
          "speed": {
            "description": "速度模式：fast|standard|deep，默认 standard",
            "enum": [
              "fast",
              "standard",
              "deep"
            ],
            "type": "string"
          },
          "steps": {
            "description": "agent 执行轨迹，逐步六维评估",
            "items": {
              "description": "agent 轨迹单步",
              "properties": {
                "agent_claim": {
                  "description": "该步 agent 的声明",
                  "type": "string"
                },
                "step": {
                  "description": "步序号，默认 1",
                  "maximum": 9007199254740991,
                  "minimum": -9007199254740991,
                  "type": "integer"
                },
                "tool": {
                  "description": "该步调用的工具名",
                  "type": "string"
                },
                "tool_input": {
                  "description": "工具输入",
                  "type": "string"
                },
                "tool_output": {
                  "description": "工具输出",
                  "type": "string"
                }
              },
              "type": "object"
            },
            "type": "array"
          },
          "task": {
            "description": "原始任务指令（task_completion 裁判用）",
            "type": "string"
          },
          "tool_schemas": {
            "additionalProperties": {},
            "description": "工具 Schema（与 trajectory 端点对齐）",
            "propertyNames": {
              "type": "string"
            },
            "type": "object"
          }
        },
        "required": [
          "agent_output"
        ],
        "type": "object"
      },
      "name": "verify_agent"
    },
    {
      "description": "逐声明幻觉核验。输入文本 → 提取声明 → 逐声明验证（supported/refuted/unverified）+ 来源引用。返回红/黄/绿汇总、每条声明的状态/置信度/理由/来源、citations、quota。耗 detect 额度（与 Web 端同源）。",
      "inputSchema": {
        "$schema": "http://json-schema.org/draft-07/schema#",
        "properties": {
          "domain": {
            "description": "领域模式：general|medical|legal|finance|education|government",
            "enum": [
              "general",
              "medical",
              "legal",
              "finance",
              "education",
              "government"
            ],
            "type": "string"
          },
          "model": {
            "description": "被测模型名（选填，分享卡片展示用）",
            "maxLength": 50,
            "type": "string"
          },
          "speed": {
            "description": "速度模式：fast|standard|deep，默认 standard",
            "enum": [
              "fast",
              "standard",
              "deep"
            ],
            "type": "string"
          },
          "strict": {
            "description": "严格模式（自我一致性），不传走全局默认",
            "type": "boolean"
          },
          "text": {
            "description": "待核验的文本",
            "maxLength": 50000,
            "minLength": 1,
            "type": "string"
          }
        },
        "required": [
          "text"
        ],
        "type": "object"
      },
      "name": "verify_text"
    }
  ]
}

What the leaf hashes

{"body":"1f5fc3016e66cc58c861c69c9bec4de346f8182773e50fac121ee8f2e0844699","id":221023,"kind":"mcp.tools","method":"probe.mcp/1","observed_us":1793027061663339,"prev":null,"subject":"476ddfbf-6081-45e9-bdce-67543d5f270f","subject_kind":"endpoint"}

Proof

Fold these into the leaf in order, hashing each sorted pair with keccak256, to reach the root of 2026-10-26 (296695 rows).
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Day root: 0xf43e8dcc3d451a4788b26be904c55e8c4c0fa75c9666796c51db38d37e2c0df8
Chained: 0x702d5ac508f08b0d1accf36af25822d8e5660367494edd17a642fb61e158a57a
Published: On Arc Testnet, block 69251483: 0x495162f8936e19f0039918feebe6c307f327cd6bf8c37219620b1970041fa9ae

How to check it · This row as JSON