← 返回
未分类 中文

Clash of Coins - Agentic Gateway

Use when an agent needs one universal entrypoint to discover and use the unified Clash of Coins gateway (sale + shop), choose the active protocol on the live...
当代理需要一个统一入口来发现并使用统一的Clash of Coins网关(销售+商店),并选择实时活动协议时使用。
clash-deployer
未分类 clawhub v1.0.5 1 版本 100000 Key: 无需
★ 0
Stars
📥 515
下载
💾 1
安装
1
版本
#latest

概述

Clash of Coins Universal Skill

Use this as the default skill for any compatible agent (OpenClaw, Claude, Cursor, Codex, and other Agent Skills clients) when the task touches Clash of Coins.

Canonical published path: /skills/clashofcoins-universal/SKILL.md

Compatibility aliases: /skills/SKILL.md, /skills/clashofcoins-universal/skill.md, /skills/clashofcoins-universal, /skills/clashofcoins/SKILL.md, /skills/clashofcoins/skill.md, /skills/clashofcoins

Scope

  • Surfaces: sale (/agentic/) and shop (/shop/)
  • Protocol model: deployment-dependent (x402, mpp, or mixed)
  • Role: universal router and discovery-first entrypoint

Functional Coverage (Sale + Shop)

  • Root discovery and routing:
  • GET /
  • GET /openapi.json
  • GET /openapi.full.json
  • GET /mcp.json
  • GET /skills/index.json
  • GET /catalog
  • check top-level x-bazaar metadata in OpenAPI for minimal Bazaar-compatible payable hints
  • GET /.well-known/x402 (compatibility metadata for enabled x402 resources; 404 when x402 is disabled)
  • GET /.well-known/mpp when MPP is enabled
  • Sale checkout coverage:
  • offers, optional quote, buy, and purchase-status polling on /agentic//*
  • request body constraints: saleId, quantity, beneficiary
  • Shop checkout coverage:
  • anonymous + recipient-scoped catalog reads on /shop/api/shop/items
  • offers, optional quote, buy, and purchase-status polling on /shop//*
  • recipient constraint: exactly one of nickname or address for quote/buy
  • Payment retry coverage:
  • x402 paid retry from latest challenge with identical method/body
  • MPP paid retry with identical JSON body and canonical mppx flow
  • Agent-wallet coverage (when enabled):
  • read live capability/funding route first: GET /agent-wallet/
  • create order (returns exact amount): POST /agent-wallet/orders
  • payable funding (default auto sweep + finalize): POST /agent-wallet//fund
  • read order state: GET /agent-wallet/orders/{orderId}
  • funding status: GET /agent-wallet//purchases/{paymentReference}
  • funding protocol is runtime/env-driven (x402 or mpp); purchase protocol in order payload can still be x402 or mpp when enabled
  • Integration coverage:
  • MCP/skills/OpenAPI consistency checks
  • scanner/registry validation via reference playbooks

Use This Skill When

  • user intent is not yet narrowed to sale or shop
  • you need one playbook to browse, route, and execute
  • you need to hand off cleanly to specialized skills after routing

Do Not Use This Skill When

  • task is explicitly sale-only and already scoped
  • task is explicitly shop-only and already scoped

Default Workflow

  1. Pick one target origin. Do not mix origins in one pass.
  2. Run discovery snapshot:
    • node scripts/discover-gateway.mjs --origin
  3. Read GET /catalog before any buy call.
  4. Route by user intent:
    • browse/compare products: stay in this skill and follow references/discovery-and-routing.md
    • buy presale/NFT lots: follow sale flow in references/purchase-playbooks.md
    • buy in-game shop goods: follow shop flow in references/purchase-playbooks.md
    • scanner/MCP/OpenAPI integration or validation: use references/integration-playbook.md
  5. Validate before execution:
    • active protocol exists on this origin
    • chosen surface matches the item (sale vs shop)
    • shop recipient rule is satisfied (exactly one of nickname or address)

Hard Rules

  • Do not mix sale and shop contracts in one purchase attempt.
  • Do not treat payment settlement as success before purchase-status/ledger confirmation.
  • Do not buy shop items from anonymous offers without recipient-scoped validation.
  • Keep unpaid and paid retry payloads identical.
  • Do not hardcode payment addresses or token/network constants outside deployment config.

Fast Defaults

  • Browse everything: GET /catalog
  • Sale buy: POST /agentic//buy
  • Shop buy: POST /shop//buy
  • Agent-wallet buy flow:
  • GET /agent-wallet/
  • POST /agent-wallet/orders
  • POST /agent-wallet//fund
  • GET /agent-wallet/orders/{orderId}
  • Sale status: GET /agentic//purchases/{paymentTx}
  • Shop status:
  • GET /shop//purchases/{paymentReference}
  • GET /shop/purchase-status/{purchaseId} (user-facing status)

Plan-Validate-Execute Loop

  1. Plan: choose one origin, one surface, and one protocol from live discovery.
  2. Validate: run the relevant checklist from references.
  3. Execute: quote or buy.
  4. Re-validate: if contract behavior differs, re-read live discovery and adjust.

Agent Wallet Simplified Flow

  1. Create order with purchase intent.
    • Read GET /agent-wallet/ and use returned fundingProtocol/fundingRoute as source of truth.
    • protocol is optional; gateway can infer a compatible protocol (for example, shop carts prefer mpp when available).
  2. Pay the funding route for returned orderId.
  3. Expect backend to auto-run sweep and finalize.
  4. Read order; if auto-execution fails, use manual recovery endpoints only as fallback.

Load References On Demand

  • Discovery and routing: references/discovery-and-routing.md
  • Purchase playbooks: references/purchase-playbooks.md
  • Integration/validator flow: references/integration-playbook.md
  • Client setup: references/client-installation.md
  • Skill eval artifacts:
  • quality evals: evals/evals.json
  • trigger evals: evals/trigger-queries.json

Output Template

### Clash of Coins Handoff

- Origin: <origin>
- Enabled protocols: <x402|mpp|both>
- Surface selected: <sale|shop>
- Why this route: <user intent + catalog evidence>
- Next endpoint: <method + path>
- Payload constraints: <recipient/beneficiary/cart rules>
- Payment retry rule: <PAYMENT-SIGNATURE or Authorization: Payment via mppx>
- Status endpoint: <exact path>
- Risks/gotchas: <if any>

版本历史

共 1 个版本

  • v1.0.5 当前
    2026-05-03 06:08 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

ai-intelligence

Self-Improving + Proactive Agent

ivangdavila
自我反思+自我批评+自我学习+自组织记忆。智能体评估自身工作、发现错误并持续改进。
★ 1,371 📥 319,668
security-compliance

Skill Vetter

spclaudehome
AI智能体技能安全预审工具。安装ClawdHub、GitHub等来源技能前,检查风险信号、权限范围及可疑模式。
★ 1,223 📥 267,323
developer-tools

Github

steipete
使用 `gh` CLI 与 GitHub 交互,通过 `gh issue`、`gh pr`、`gh run` 和 `gh api` 管理议题、PR、CI 运行及高级查询。
★ 672 📥 324,946