MSG Chain 能否开发类似 OKX AI（okx.ai）项目分析

数据来源：MSG Chain 代码库核实、okx.ai 官方

主网状态: No-Go — 当前 MSGChain 主网裁决为 No-Go，以下内容反映代码实际状态，不代表生产可用。

一、OKX AI 核心能力拆解

OKX AI（okx.ai）是全球首个 A2A（Agent-to-Agent）经济体平台，核心功能包括：

二、MSG Chain 现有能力覆盖度

✅ 已有或部分实现

❌ 缺失或未实现

三、缺失模块完整实现方案

以下逐一给出每个缺失模块的完整设计方案，包括 Schema、状态模型、核心逻辑和集成方式。

模块 1：task_marketplace_v1 — 任务市场合约

对应 OKX AI 的「任务发布 → Agent 竞标 → 匹配执行」

状态模型

// --- 任务状态枚举 ---
pub enum TaskStatus {
    Open,           // 开放竞标
    InProgress,     // 已匹配，执行中
    Delivered,      // 已交付，待确认
    Completed,      // 已完成（付款已释放）
    Cancelled,      // 已取消
    Disputed,       // 争议中
}

// --- 任务数据结构 ---
pub struct Task {
    pub id: String,                    // 唯一任务 ID
    pub title: String,                 // 任务标题
    pub description: String,           // 任务描述
    pub requester: Addr,               // 发布者地址
    pub budget: Uint128,               // 预算金额
    pub asset: String,                 // 支付资产（cw20 合约地址或 "native"）
    pub deadline: u64,                 // 截止时间戳
    pub required_capabilities: Vec<String>, // 所需 Agent 能力
    pub status: TaskStatus,           // 当前状态
    pub assigned_agent: Option<Addr>,  // 中标的 Agent
    pub escrow_payment_id: String,    // 关联的 escrow payment ID
    pub created_at: u64,
    pub updated_at: u64,
}

// --- 竞标数据结构 ---
pub struct Bid {
    pub id: String,
    pub task_id: String,
    pub agent_id: String,             // Agent 的 DID/registry ID
    pub bidder: Addr,                 // Agent 所有者地址
    pub amount: Uint128,              // 竞标价格
    pub proposal: String,             // 执行方案描述
    pub estimated_delivery: u64,       // 预计交付时间
    pub status: BidStatus,            // pending / accepted / rejected
    pub created_at: u64,
}

pub enum BidStatus {
    Pending,
    Accepted,
    Rejected,
}

ExecuteMsg Schema

{
  "title": "ExecuteMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["create_task"],
      "properties": {
        "create_task": {
          "type": "object",
          "required": [
            "task_id", "title", "description", "budget", "asset",
            "deadline", "required_capabilities"
          ],
          "properties": {
            "task_id": {"type": "string"},
            "title": {"type": "string"},
            "description": {"type": "string"},
            "budget": {"type": "string"},
            "asset": {"type": "string"},
            "deadline": {"type": "integer", "format": "uint64"},
            "required_capabilities": {
              "type": "array", "items": {"type": "string"}
            }
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["submit_bid"],
      "properties": {
        "submit_bid": {
          "type": "object",
          "required": [
            "bid_id", "task_id", "agent_id", "amount", "proposal", "estimated_delivery"
          ],
          "properties": {
            "bid_id": {"type": "string"},
            "task_id": {"type": "string"},
            "agent_id": {"type": "string"},
            "amount": {"type": "string"},
            "proposal": {"type": "string"},
            "estimated_delivery": {"type": "integer", "format": "uint64"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["accept_bid"],
      "properties": {
        "accept_bid": {
          "type": "object",
          "required": ["task_id", "bid_id"],
          "properties": {
            "task_id": {"type": "string"},
            "bid_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["confirm_delivery"],
      "properties": {
        "confirm_delivery": {
          "type": "object",
          "required": ["task_id", "delivery_proof_hash"],
          "properties": {
            "task_id": {"type": "string"},
            "delivery_proof_hash": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["raise_dispute"],
      "properties": {
        "raise_dispute": {
          "type": "object",
          "required": ["task_id", "reason"],
          "properties": {
            "task_id": {"type": "string"},
            "reason": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["cancel_task"],
      "properties": {
        "cancel_task": {
          "type": "object",
          "required": ["task_id", "reason"],
          "properties": {
            "task_id": {"type": "string"},
            "reason": {"type": "string"}
          }
        }
      }
    }
  ]
}

QueryMsg Schema

{
  "title": "QueryMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["get_task"],
      "properties": {
        "get_task": {
          "type": "object",
          "required": ["task_id"],
          "properties": {"task_id": {"type": "string"}}
        }
      }
    },
    {
      "type": "object",
      "required": ["list_tasks"],
      "properties": {
        "list_tasks": {
          "type": "object",
          "properties": {
            "status": {"type": ["string", "null"]},
            "requester": {"type": ["string", "null"]},
            "capability": {"type": ["string", "null"]},
            "limit": {"type": ["integer", "null"], "format": "uint32"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["list_bids"],
      "properties": {
        "list_bids": {
          "type": "object",
          "required": ["task_id"],
          "properties": {"task_id": {"type": "string"}}
        }
      }
    },
    {
      "type": "object",
      "required": ["get_bid"],
      "properties": {
        "get_bid": {
          "type": "object",
          "required": ["bid_id"],
          "properties": {"bid_id": {"type": "string"}}
        }
      }
    }
  ]
}

核心业务逻辑

create_task:
  1. 验证 task_id 唯一性
  2. 验证 budget > 0 且 deadline > now
  3. 将 budget 转入合约 escrow（通过 CW20 transfer_from 或 native coin）
  4. 创建 Task 记录，状态 = Open
  5. 通过 Agent API 触发事件通知

submit_bid:
  1. 验证 task 存在且状态 = Open
  2. 验证 agent_id 在 agent_registry_v1 中已注册
  3. 验证 deadline 未过期
  4. 创建 Bid 记录，状态 = Pending

accept_bid:
  1. 仅 task.requester 可调用
  2. 验证 task 状态 = Open
  3. 验证 bid 状态 = Pending
  4. 将 task.status 设为 InProgress
  5. 将 bid.status 设为 Accepted（其余 bid 设为 Rejected）
  6. 调用 agent_payment_v1 创建 escrow payment

confirm_delivery:
  1. 仅 task.requester 可调用
  2. 验证 task 状态 = InProgress
  3. 验证 delivery_proof_hash 有效
  4. 将 task.status 设为 Completed
  5. 调用 agent_payment_v1.settle_payment 释放资金给 Agent

raise_dispute:
  1. requester 或 assigned_agent 均可调用
  2. 验证 task 状态 = InProgress 或 Delivered
  3. 将 task.status 设为 Disputed
  4. 创建 DisputeRecord，调用 dispute_resolution_v1

cancel_task:
  1. 仅 requester 可调用
  2. 验证 task 状态 = Open
  3. 退回 escrow 资金给 requester
  4. 将 task.status 设为 Cancelled

与 MSG Chain 现有合约的集成

模块 2：dispute_resolution_v1 — 争议仲裁合约

对应 OKX AI 的「评估人系统：质押 → 争议投票 → 裁决 → 罚没」

状态模型

pub enum DisputeStatus {
    Open,           // 等待投票
    Voting,         // 投票进行中
    Resolved,       // 已裁决
    Appealed,       // 已上诉
}

pub enum Verdict {
    ForRequester,   // 支持发布者
    ForAgent,       // 支持 Agent
    Split(Fraction),// 按比例分配
}

pub struct Dispute {
    pub id: String,
    pub task_id: String,
    pub initiator: Addr,         // 发起方
    pub opponent: Addr,          // 被诉方
    pub reason: String,          // 争议原因
    pub evidence_hash: String,   // 证据哈希
    pub status: DisputeStatus,
    pub escrow_amount: Uint128,  // 争议涉及的金额
    pub verdict: Option<Verdict>,
    pub created_at: u64,
    pub resolved_at: Option<u64>,
}

pub struct Evaluator {
    pub address: Addr,
    pub staked_amount: Uint128,
    pub reputation_score: u64,   // 历史裁决准确率
    pub total_votes: u64,
    pub correct_votes: u64,
    pub joined_at: u64,
}

pub struct Vote {
    pub dispute_id: String,
    pub evaluator: Addr,
    pub verdict: Verdict,
    pub rationale_hash: String,  // 裁决理由哈希
    pub voted_at: u64,
    pub weight: u64,             // 投票权重（基于质押量）
}

ExecuteMsg Schema

{
  "title": "ExecuteMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["register_evaluator"],
      "properties": {
        "register_evaluator": {
          "type": "object",
          "required": ["stake_amount"],
          "properties": {
            "stake_amount": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["unregister_evaluator"],
      "properties": {
        "unregister_evaluator": {
          "type": "object",
          "properties": {}
        }
      }
    },
    {
      "type": "object",
      "required": ["create_dispute"],
      "properties": {
        "create_dispute": {
          "type": "object",
          "required": ["dispute_id", "task_id", "opponent", "reason", "evidence_hash", "escrow_amount"],
          "properties": {
            "dispute_id": {"type": "string"},
            "task_id": {"type": "string"},
            "opponent": {"type": "string"},
            "reason": {"type": "string"},
            "evidence_hash": {"type": "string"},
            "escrow_amount": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["cast_vote"],
      "properties": {
        "cast_vote": {
          "type": "object",
          "required": ["dispute_id", "verdict", "rationale_hash"],
          "properties": {
            "dispute_id": {"type": "string"},
            "verdict": {
              "type": "string",
              "enum": ["for_requester", "for_agent", "split"]
            },
            "split_ratio": {
              "type": ["string", "null"]
            },
            "rationale_hash": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["resolve_dispute"],
      "properties": {
        "resolve_dispute": {
          "type": "object",
          "required": ["dispute_id"],
          "properties": {
            "dispute_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["slash_evaluator"],
      "properties": {
        "slash_evaluator": {
          "type": "object",
          "required": ["evaluator", "dispute_id", "reason"],
          "properties": {
            "evaluator": {"type": "string"},
            "dispute_id": {"type": "string"},
            "reason": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["withdraw_stake"],
      "properties": {
        "withdraw_stake": {
          "type": "object",
          "properties": {}
        }
      }
    }
  ]
}

核心业务逻辑

register_evaluator:
  1. 将 staked_amount 从调用者转入合约（CW20 transfer_from）
  2. 创建或更新 Evaluator 记录
  3. 发出 EvaluatorRegistered 事件

create_dispute:
  1. 验证 dispute_id 唯一
  2. 验证 escrow_amount 已在 task_marketplace_v1 中锁定
  3. 创建 Dispute 记录，状态 = Open
  4. 随机选择 N 个评估人（按质押权重采样）
  5. 发出 DisputeCreated 事件

cast_vote:
  1. 仅已注册的评估人可调用
  2. 验证投票期未结束（投票窗口 = 24h）
  3. 每个 evaluator 每个 dispute 仅一票
  4. 记录投票，权重 = evaluator.staked_amount
  5. 如果投票数 >= 阈值（如 5 票）→ 自动 resolve

resolve_dispute:
  1. 统计所有投票的加权结果
  2. 根据多数决确定 Verdict
  3. 通知 task_marketplace_v1 执行裁决结果（释放资金/退款/按比例分配）
  4. 更新 Dispute 状态为 Resolved
  5. 奖励与多数决一致的评估人，标记少数的供后续 slash

slash_evaluator:
  1. 需 DAO 提案或系统自动触发
  2. 确认该 evaluator 在特定 dispute 中投了恶意票
  3. 罚没部分 stake 转入 treasury
  4. 降低其 reputation_score

与 MSG Chain 现有合约的集成

模块 3：agent_reputation_v1 — Agent 声誉合约

对应 OKX AI 的「Agent 链上声誉积累」

状态模型

pub struct AgentReputation {
    pub agent_id: String,           // 对应 agent_registry_v1 的 agent_id
    pub aidid: String,              // 对应 DID
    pub total_tasks_completed: u64,
    pub total_tasks_failed: u64,
    pub total_earned: Uint128,
    pub average_rating: f64,        // 1.0 - 5.0
    pub rating_count: u64,
    pub dispute_win_rate: f64,      // 争议胜率
    pub response_time_avg: u64,     // 平均响应时间（秒）
    pub last_activity: u64,
    pub badges: Vec<String>,        // 徽章列表
}

pub struct Rating {
    pub rating_id: String,
    pub task_id: String,
    pub rater: Addr,
    pub score: u8,                  // 1-5
    pub comment_hash: String,
    pub created_at: u64,
}

ExecuteMsg Schema

{
  "title": "ExecuteMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["submit_rating"],
      "properties": {
        "submit_rating": {
          "type": "object",
          "required": ["rating_id", "task_id", "agent_id", "score", "comment_hash"],
          "properties": {
            "rating_id": {"type": "string"},
            "task_id": {"type": "string"},
            "agent_id": {"type": "string"},
            "score": {"type": "integer", "minimum": 1, "maximum": 5},
            "comment_hash": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["record_task_completion"],
      "properties": {
        "record_task_completion": {
          "type": "object",
          "required": ["agent_id", "task_id", "earned_amount"],
          "properties": {
            "agent_id": {"type": "string"},
            "task_id": {"type": "string"},
            "earned_amount": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["record_task_failure"],
      "properties": {
        "record_task_failure": {
          "type": "object",
          "required": ["agent_id", "task_id", "reason"],
          "properties": {
            "agent_id": {"type": "string"},
            "task_id": {"type": "string"},
            "reason": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["record_dispute_outcome"],
      "properties": {
        "record_dispute_outcome": {
          "type": "object",
          "required": ["agent_id", "dispute_id", "won"],
          "properties": {
            "agent_id": {"type": "string"},
            "dispute_id": {"type": "string"},
            "won": {"type": "boolean"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["award_badge"],
      "properties": {
        "award_badge": {
          "type": "object",
          "required": ["agent_id", "badge_name"],
          "properties": {
            "agent_id": {"type": "string"},
            "badge_name": {"type": "string"}
          }
        }
      }
    }
  ]
}

查询接口

{
  "title": "QueryMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["get_reputation"],
      "properties": {
        "get_reputation": {
          "type": "object",
          "required": ["agent_id"],
          "properties": {"agent_id": {"type": "string"}}
        }
      }
    },
    {
      "type": "object",
      "required": ["top_agents"],
      "properties": {
        "top_agents": {
          "type": "object",
          "properties": {
            "sort_by": {
              "type": "string",
              "enum": ["rating", "tasks_completed", "total_earned"]
            },
            "limit": {"type": "integer", "format": "uint32"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["list_ratings"],
      "properties": {
        "list_ratings": {
          "type": "object",
          "required": ["agent_id"],
          "properties": {
            "agent_id": {"type": "string"},
            "limit": {"type": ["integer", "null"], "format": "uint32"}
          }
        }
      }
    }
  ]
}

徽章系统设计

模块 4：stablecoin_v1 — 稳定币合约

对应 OKX AI 的「稳定币即时结算」

实现方案

基于 msg_token_cw20 的模式，新建一个符合 CW20 标准的稳定币合约。

// 使用 CW20 基础标准 + 扩展
//
// InstantiateMsg:
//   - name: "MSG USD"
//   - symbol: "USDM"
//   - decimals: 6
//   - initial_balances: []
//   - mint: { minter: "{treasury_address}" }
//
// 扩展方法：
//   - mint (仅合约 owner/多签可调用)
//   - burn (仅合约 owner/多签可调用)
//   - pause / unpause (紧急暂停)

无需新合约的替代方案

如果 MSG Chain 已有 msg_token_cw20，也可以直接使用它作为支付资产，不另发稳定币。OKX AI 的核心价值不在于「自己的稳定币」，而在于「稳定币结算通道」。因此：

方案 A：使用 msg_token_cw20 作为原生支付资产
方案 B：新建 stablecoin_v1 作为 USD 锚定币（需外部预言机喂价）
方案 C：跨链桥接 USDC/USDT（需 bridge_adapter_v1 就绪）

推荐先采用 方案 A 快速构建，等 bridge 上线后再逐步迁移到真实稳定币。

模块 5：agentic_wallet_v1 — Agent 自主钱包合约

对应 OKX AI 的「AI Agent 自主管理的链上钱包」

状态模型

pub struct AgenticWallet {
    pub wallet_id: String,
    pub agent_id: String,           // 关联的 Agent
    pub owner: Addr,                // 人类所有者（可 override）
    pub balances: Map<String, Uint128>, // asset -> balance
    pub spending_limits: Map<String, SpendingLimit>, // per-asset 限制
    pub timelock_delay: u64,        // 执行延迟（秒）
    pub authorized_actions: Vec<String>, // 允许的 action 列表
    pub co_signer_required: bool,   // 是否需要共同签名
    pub co_signer: Option<Addr>,    // 共同签名人
    pub created_at: u64,
}

pub struct SpendingLimit {
    pub max_per_tx: Uint128,        // 单笔最大金额
    pub max_daily: Uint128,         // 日累计上限
    pub daily_spent: Uint128,       // 当日已花费
    pub last_reset: u64,            // 每日重置时间
}

pub struct PendingExecution {
    pub execution_id: String,
    pub wallet_id: String,
    pub action: WalletAction,
    pub amount: Uint128,
    pub asset: String,
    pub target: Addr,
    pub data: Option<Binary>,       // 合约调用数据
    pub submitted_at: u64,
    pub execute_after: u64,         // timelock 解锁时间
    pub executed: bool,
    pub cancelled: bool,
}

ExecuteMsg Schema

{
  "title": "ExecuteMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["create_wallet"],
      "properties": {
        "create_wallet": {
          "type": "object",
          "required": ["wallet_id", "agent_id"],
          "properties": {
            "wallet_id": {"type": "string"},
            "agent_id": {"type": "string"},
            "timelock_delay": {"type": ["integer", "null"], "format": "uint64"},
            "spending_limits": {
              "type": ["array", "null"],
              "items": {
                "type": "object",
                "required": ["asset", "max_per_tx", "max_daily"],
                "properties": {
                  "asset": {"type": "string"},
                  "max_per_tx": {"type": "string"},
                  "max_daily": {"type": "string"}
                }
              }
            },
            "co_signer": {"type": ["string", "null"]}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["submit_execution"],
      "properties": {
        "submit_execution": {
          "type": "object",
          "required": ["execution_id", "wallet_id", "action", "amount", "asset", "target"],
          "properties": {
            "execution_id": {"type": "string"},
            "wallet_id": {"type": "string"},
            "action": {
              "type": "string",
              "enum": ["transfer", "contract_execute", "payment"]
            },
            "amount": {"type": "string"},
            "asset": {"type": "string"},
            "target": {"type": "string"},
            "data": {"type": ["string", "null"]}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["co_sign_execution"],
      "properties": {
        "co_sign_execution": {
          "type": "object",
          "required": ["execution_id"],
          "properties": {
            "execution_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["execute_pending"],
      "properties": {
        "execute_pending": {
          "type": "object",
          "required": ["execution_id"],
          "properties": {
            "execution_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["cancel_pending"],
      "properties": {
        "cancel_pending": {
          "type": "object",
          "required": ["execution_id"],
          "properties": {
            "execution_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["update_limits"],
      "properties": {
        "update_limits": {
          "type": "object",
          "required": ["wallet_id", "asset", "max_per_tx", "max_daily"],
          "properties": {
            "wallet_id": {"type": "string"},
            "asset": {"type": "string"},
            "max_per_tx": {"type": "string"},
            "max_daily": {"type": "string"}
          }
        }
      }
    }
  ]
}

核心业务逻辑

create_wallet:
  1. 验证 agent_id 在 agent_registry_v1 中已注册
  2. 验证 wallet_id 唯一
  3. 创建 AgenticWallet 记录
  4. 初始存入少量 MSG 作为 gas 费用

submit_execution (由 AI Agent 通过 Agent API 调用):
  1. 验证 wallet 存在且 agent_id 匹配
  2. 验证 action 在 authorized_actions 中
  3. 验证 spending_limits（单笔 + 日累计）
  4. 如果 co_signer_required → 设置 execute_after = now + timelock_delay
  5. 否则 → execute_after = now
  6. 创建 PendingExecution 记录
  7. 通过 Agent API 事件通知 owner

execute_pending:
  1. 验证 now >= execute_after
  2. 验证未执行且未取消
  3. 执行具体 action（transfer/contract_execute/payment）
  4. 标记为 executed
  5. 更新 spending_limits.daily_spent

与 Agent API 的集成

# AI Agent 通过 Agent API 调用示例
POST /agent/v1/wallet/execution
{
  "wallet_id": "wallet_agent_001",
  "action": "payment",
  "amount": "1000000",  // 1 USDM (18 decimals)
  "asset": "usdm",
  "target": "msg1agent_service_provider_address"
}

# Agent API 的 wrapProtected 中间件会验证 API Key
# 然后调用 agentic_wallet_v1 合约的 submit_execution

模块 6：skill_registry_v1 — 技能注册市场合约

对应 OKX AI 的「Skills Marketplace」

状态模型

pub struct Skill {
    pub skill_id: String,
    pub name: String,
    pub description: String,
    pub publisher: Addr,
    pub version: String,
    pub category: String,            // "trading", "data", "analysis", "automation"
    pub price: Uint128,              // 使用价格
    pub price_type: PriceType,       // OneTime / Subscription / PerUse
    pub execution_hash: String,      // 可执行代码哈希（IPFS 或链下引用）
    pub schema_url: String,          // 输入/输出 Schema
    pub rating: f64,
    pub install_count: u64,
    pub verified: bool,              // 是否经过安全审核
    pub created_at: u64,
}

pub struct Installation {
    pub install_id: String,
    pub skill_id: String,
    pub installer: Addr,             // 安装者 Agent ID 或用户地址
    pub agent_id: String,
    pub installed_at: u64,
    pub active: bool,
}

ExecuteMsg Schema

{
  "title": "ExecuteMsg",
  "oneOf": [
    {
      "type": "object",
      "required": ["publish_skill"],
      "properties": {
        "publish_skill": {
          "type": "object",
          "required": [
            "skill_id", "name", "description", "version", "category",
            "price", "price_type", "execution_hash"
          ],
          "properties": {
            "skill_id": {"type": "string"},
            "name": {"type": "string"},
            "description": {"type": "string"},
            "version": {"type": "string"},
            "category": {"type": "string"},
            "price": {"type": "string"},
            "price_type": {"type": "string"},
            "execution_hash": {"type": "string"},
            "schema_url": {"type": ["string", "null"]}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["install_skill"],
      "properties": {
        "install_skill": {
          "type": "object",
          "required": ["skill_id", "agent_id"],
          "properties": {
            "skill_id": {"type": "string"},
            "agent_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["uninstall_skill"],
      "properties": {
        "uninstall_skill": {
          "type": "object",
          "required": ["skill_id", "agent_id"],
          "properties": {
            "skill_id": {"type": "string"},
            "agent_id": {"type": "string"}
          }
        }
      }
    },
    {
      "type": "object",
      "required": ["verify_skill"],
      "properties": {
        "verify_skill": {
          "type": "object",
          "required": ["skill_id", "verified"],
          "properties": {
            "skill_id": {"type": "string"},
            "verified": {"type": "boolean"}
          }
        }
      }
    }
  ]
}

与前端 Skills 市场的集成

前端 Skills 市场 → skill_registry_v1 合约查询 → 展示可用技能
Agent 选择技能 → install_skill → 合约记录安装
Agent 执行任务时 → 引用已安装的 skill_id → 验证执行权限

模块 7：前端 dApp — 任务市场 + 争议面板

技术栈

框架: React + TypeScript + Vite
钱包: Keplr / MSG Chain Wallet（CosmWasm 兼容）
合约交互: cosmjs (CosmWasmClient / SigningCosmWasmClient)
AI 集成: MCP Client 或直接 Agent API 调用
状态管理: React Query (TanStack Query)
UI: Tailwind CSS + shadcn/ui

页面路由设计

核心 API 调用模式

// 1. 连接钱包
import { CosmWasmClient } from "@cosmjs/cosmwasm-stargate";

const client = await CosmWasmClient.connect("http://localhost:26657");

// 2. 查询任务市场
const tasks = await client.queryContractSmart(
  taskMarketplaceAddress,
  { list_tasks: { status: "open", limit: 20 } }
);

// 3. 提交竞标（通过签名交易）
import { SigningCosmWasmClient } from "@cosmjs/cosmwasm-stargate";
import { DirectSecp256k1HdWallet } from "@cosmjs/proto-signing";

const signingClient = await SigningCosmWasmClient.connectWithSigner(
  rpcEndpoint,
  wallet
);

const result = await signingClient.execute(
  senderAddress,
  taskMarketplaceAddress,
  {
    submit_bid: {
      bid_id: uuid(),
      task_id: taskId,
      agent_id: agentId,
      amount: "5000000",
      proposal: "I will complete this task in 2 hours...",
      estimated_delivery: Math.floor(Date.now() / 1000) + 7200,
    }
  },
  "auto"
);

// 4. 通过 Agent API 调用（AI Agent 使用）
const response = await fetch("https://msgchain.org/agent/v1/query/account/msg1...", {
  headers: { "X-API-Key": agentApiKey }
});
const data = await response.json();

Agent Marketplace 前端组件清单

components/
├── layout/
│   ├── Header.tsx          // 导航、钱包连接、搜索
│   ├── Footer.tsx
│   └── Sidebar.tsx
├── task/
│   ├── TaskCard.tsx        // 任务卡片（列表展示）
│   ├── TaskDetail.tsx      // 任务详情
│   ├── TaskForm.tsx        // 创建任务表单
│   ├── BidList.tsx         // 竞标列表
│   ├── BidForm.tsx         // 提交竞标
│   └── TaskTimeline.tsx    // 任务状态时间线
├── agent/
│   ├── AgentCard.tsx       // Agent 卡片
│   ├── AgentDetail.tsx     // Agent 详情（声誉、技能、评价）
│   ├── AgentSearch.tsx     // Agent 搜索过滤
│   └── ReputationBadge.tsx // 声誉徽章展示
├── dispute/
│   ├── DisputeList.tsx     // 争议列表
│   ├── DisputeDetail.tsx   // 争议详情 + 投票
│   ├── VotePanel.tsx       // 投票面板
│   └── EvaluatorPanel.tsx  // 评估人面板
├── wallet/
│   ├── WalletDashboard.tsx // Agentic Wallet 仪表盘
│   ├── ExecutionQueue.tsx  // 待执行队列
│   ├── LimitSettings.tsx   // 限额设置
│   └── TransactionHistory.tsx
├── skills/
│   ├── SkillCard.tsx
│   ├── SkillDetail.tsx
│   └── SkillInstaller.tsx
└── shared/
    ├── WalletConnector.tsx
    ├── AgentAPIKeys.tsx
    └── NetworkStatus.tsx

模块 8：MCP Server — 桥接 AI Agent 与 MSG Chain

对应 OKX AI 的「OnchainOS + MCP 支持」，让 AI Agent 通过自然语言直接操作链

MCP Server 实现

// mcp-server/src/index.ts
// 基于 Model Context Protocol 标准

import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import {
  CallToolRequestSchema,
  ListToolsRequestSchema,
} from "@modelcontextprotocol/sdk/types.js";

const server = new Server(
  { name: "msgchain-agent-mcp", version: "1.0.0" },
  { capabilities: { tools: {} } }
);

// 注册工具
server.setRequestHandler(ListToolsRequestSchema, async () => ({
  tools: [
    {
      name: "query_task_market",
      description: "查询 MSG Chain 任务市场",
      inputSchema: {
        type: "object",
        properties: {
          status: { type: "string", enum: ["open", "in_progress", "completed"] },
          limit: { type: "number" }
        }
      }
    },
    {
      name: "submit_bid",
      description: "AI Agent 对任务提交竞标",
      inputSchema: {
        type: "object",
        properties: {
          task_id: { type: "string" },
          agent_id: { type: "string" },
          amount: { type: "string" },
          proposal: { type: "string" }
        },
        required: ["task_id", "agent_id", "amount"]
      }
    },
    {
      name: "check_reputation",
      description: "查询 Agent 链上声誉",
      inputSchema: {
        type: "object",
        properties: {
          agent_id: { type: "string" }
        },
        required: ["agent_id"]
      }
    },
    {
      name: "wallet_execute",
      description: "通过 Agentic Wallet 执行链上操作",
      inputSchema: {
        type: "object",
        properties: {
          wallet_id: { type: "string" },
          action: { type: "string", enum: ["transfer", "payment"] },
          amount: { type: "string" },
          asset: { type: "string" },
          target: { type: "string" }
        },
        required: ["wallet_id", "action", "amount", "asset", "target"]
      }
    },
    {
      name: "list_skills",
      description: "列出可安装的技能",
      inputSchema: {
        type: "object",
        properties: {
          category: { type: "string" },
          limit: { type: "number" }
        }
      }
    },
    {
      name: "install_skill",
      description: "为当前 Agent 安装技能",
      inputSchema: {
        type: "object",
        properties: {
          skill_id: { type: "string" },
          agent_id: { type: "string" }
        },
        required: ["skill_id", "agent_id"]
      }
    },
    {
      name: "resolve_dispute",
      description: "评估人对争议投票裁决",
      inputSchema: {
        type: "object",
        properties: {
          dispute_id: { type: "string" },
          verdict: { type: "string", enum: ["for_requester", "for_agent", "split"] },
          rationale: { type: "string" }
        },
        required: ["dispute_id", "verdict"]
      }
    }
  ]
}));

// 工具执行处理
server.setRequestHandler(CallToolRequestSchema, async (request) => {
  const { name, arguments: args } = request.params;

  switch (name) {
    case "query_task_market": {
      // 调用 task_marketplace_v1 合约的 Query
      const tasks = await cosmwasmClient.queryContractSmart(
        taskMarketplaceAddr,
        { list_tasks: { status: args.status, limit: args.limit || 20 } }
      );
      return { content: [{ type: "text", text: JSON.stringify(tasks) }] };
    }
    case "submit_bid": {
      // 通过 SigningCosmWasmClient 签名并提交交易
      const tx = await signingClient.execute(
        senderAddr,
        taskMarketplaceAddr,
        { submit_bid: { ...args, bid_id: crypto.randomUUID(), estimated_delivery: 0 } },
        "auto"
      );
      return { content: [{ type: "text", text: JSON.stringify(tx) }] };
    }
    // ... 其他工具类似实现
  }
});

const transport = new StdioServerTransport();
await server.connect(transport);

使用方式

# AI Agent 通过 MCP 连接
npx install @msg-chain/mcp-server

# 在 Claude Code / Cursor / OpenClaw 中配置
# mcp_servers:
#   msgchain:
#     command: npx
#     args: ["@msg-chain/mcp-server"]
#     env:
#       MSG_RPC_URL: "http://localhost:26657"
#       MSG_AGENT_API_KEY: "your-agent-api-key"

# AI 开发者直接通过自然语言操作：
# "显示任务市场上所有 open 状态的任务"
# "对任务 task_001 提交 5 MSG 的竞标"
# "查询 agent_alice 的链上声誉"

四、完整开发工作流

步骤 1：初始化项目结构

mkdir msg-chain-okx-ai && cd msg-chain-okx-ai

# 合约目录
mkdir -p contracts/
cd contracts/

# 每个合约独立的 CosmWasm 项目
cargo generate --git https://github.com/CosmWasm/cw-template.git --name task_marketplace_v1
cargo generate --git https://github.com/CosmWasm/cw-template.git --name dispute_resolution_v1
cargo generate --git https://github.com/CosmWasm/cw-template.git --name agent_reputation_v1
cargo generate --git https://github.com/CosmWasm/cw-template.git --name stablecoin_v1
cargo generate --git https://github.com/CosmWasm/cw-template.git --name agentic_wallet_v1
cargo generate --git https://github.com/CosmWasm/cw-template.git --name skill_registry_v1

# 前端目录
cd ..
mkdir -p frontend && cd frontend
npm create vite@latest . -- --template react-ts
npm install @cosmjs/cosmwasm-stargate @cosmjs/proto-signing @cosmjs/encoding
npm install @tanstack/react-query tailwindcss shadcn/ui

# MCP Server
cd ..
mkdir -p mcp-server && cd mcp-server
npm init -y
npm install @modelcontextprotocol/sdk typescript

步骤 2：按顺序开发合约

第一周: agent_reputation_v1 + stablecoin_v1（依赖最少，快速验证）
第二周: agentic_wallet_v1（依赖 agent_reputation_v1 的身份验证）
第三周: task_marketplace_v1（核心市场逻辑，依赖上述所有）
第四周: dispute_resolution_v1（争议仲裁，依赖 task_marketplace_v1）
第五周: skill_registry_v1（技能市场，独立）
第六周: 集成测试 + E2E 测试

步骤 3：前端开发（与合约开发并行）

第七周: 基础框架 + 钱包连接 + Agent/任务浏览
第八周: 任务发布/竞标/管理功能
第九周: Agent 详情/声誉展示 + 争议面板
第十周: Agentic Wallet 管理 + Skills 市场

步骤 4：MCP Server + AI 集成

第十一周: MCP Server 开发
第十二周: AI Agent 端到端测试（自然语言 → 任务 → 执行 → 结付）

步骤 5：测试与部署

第十三周: 本地 E2E 测试（基于 MSG Chain 本地沙箱）
第十四周: 安全审计 + 文档 + 部署到 testnet（等待 MSG Chain 公开网络）

五、AI Agent 开发提示词模板

可将其输入到任何 AI coding agent（Claude Code / Cursor / OpenClaw）中，用于复刻开发本项目：

你是一个 AI 开发代理，目标是在 MSG Chain 区块链上复刻类似 OKX AI 的 Agent-to-Agent 经济体平台。

## 技术栈
- 智能合约: Rust + CosmWasm
- 前端: React + TypeScript + Vite + cosmjs
- AI 集成: MCP (Model Context Protocol)
- 共识算法: Round-Robin + DAR
- 签名算法: Dilithium-5（后量子安全）

## 已有基础设施（可直接集成）
- agent_registry_v1: Agent 注册与发现（参考 schema 详见白皮书）
- agent_payment_v1: Agent 支付意图与结算（参考 schema 详见白皮书）
- aidid_did_registry_v1: DID 身份注册（参考 schema 详见白皮书）
- ai_agent_constitution_v1: Agent 行为宪章（参考 schema 详见白皮书）
- dao_governance_v1: DAO 治理提案/投票/执行
- foundation_treasury_v2: 基金会金库多签
- msg_token_cw20: 原生 CW20 代币
- Agent API: /agent/v1/query/*（只读）+ /agent/v1/wallet/*（受保护写路径）

## 需要新建的合约
1. task_marketplace_v1 - 任务发布/竞标/匹配/交付确认
2. dispute_resolution_v1 - 评估人质押/争议投票/罚没
3. agent_reputation_v1 - Agent 链上声誉评分
4. stablecoin_v1 - CW20 标准稳定币
5. agentic_wallet_v1 - AI 自主管理钱包
6. skill_registry_v1 - 技能模块市场

## 开发入口
- Agent Entry: https://msgchain.org/whitepaper/agent_entry.json
- Developer Entry: https://msgchain.org/whitepaper/developer_entry.json
- 合约 Reference: https://msgchain.org/whitepaper/contract_reference/core_contracts.json
- Agent API OpenAPI: https://msgchain.org/whitepaper/api_specs/openapi/agent_surface.yaml
- 外部 AI Agent 引导提示词: https://msgchain.org/whitepaper/integration_examples/external_ai_agent_bootstrap_prompt.md

## 实现原则
1. 所有合约遵循 CosmWasm 标准模式（InstantiateMsg / ExecuteMsg / QueryMsg）
2. 每个合约有完整的 schema 和 E2E 测试
3. 参考现有 MSG 合约的边界模式（implemented / partial / stub 标注）
4. 高风险操作保留人工审批门禁（DAO 提案或金库多签）
5. 利用 Agent API 的公共查询端点和受保护写端点
6. 使用 MCP Server 桥接 AI Agent 与链上合约

六、可行性结论

🟢 技术上：完全可实现

MSG Chain 已有的 agent_registry_v1、agent_payment_v1、aidid_did_registry_v1、ai_agent_constitution_v1 等合约已经提供了坚实的底层基础设施。缺失的 6 个合约都是标准的 CosmWasm 业务合约，可独立开发，与公链核心无关。

🟡 当前阶段：可本地开发，不可生产部署

唯一不可控的外部依赖是「MSG Chain 公开 testnet/mainnet 尚未上线」。所有代码可在本地沙箱中完整运行和测试。

📊 开发工作量估算

七、差异化策略

利用 MSG Chain 的独特优势，建议聚焦以下方向：

核心差异化：后量子安全的 Agent 经济体

MSG Chain 的 Dilithium-5 + ChainID=1 提供了对量子计算攻击的原生防御
→ 适合对安全性要求极高的 Agent 金融场景
→ 例如：高价值 Agent 支付、军事/政务级 Agent 协作、长期资产托管

不建议复制的方向

• ❌ AI 交易机器人/策略市场（MSG 不是交易所，缺乏流动性深度）
• ❌ 跨链聚合器（bridge 尚在规划中，且 EVM 链已有成熟方案）
• ❌ CEX 级稳定币出入金（需要交易所牌照和银行通道）

八、开发资源索引

MSG Chain 已有合约参考

推荐开发路径

developer_entry.json
  → recommended_contract_bootstrap_order
    → contract.html（合约运行时）
    → registry.html（寻址层）
    → rpc.html（API 接口）
    → recipes/contract_minimal.json（Recipe 模板）
    → 逐个实现 6 个新合约
    → 前端 dApp
    → MCP Server
    → E2E 测试