Token导航 LogoToken导航

Agent Skill · 研究检索

masumi(masumi 搜索)

masumi 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

masumi 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 56,419。

总安装 451周安装 19GitHub Stars 1下载 158核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:masumi(masumi 搜索)
来源仓库:https://github.com/masumi-network/masumi-skills
仓库路径:skills/masumi
安装命令:
npx skills add https://github.com/masumi-network/masumi-skills --skill masumi
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/masumi-network/masumi-skills --skill masumi

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

masumi 用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词、任务场景或来源线索快速定位候选结果时使用。通过 npx skills add 命令从指定 GitHub 仓库安装,需确认权限和维护状态。使用前建议核实是否会触发联网、命令执行或文件读写操作。

masumi 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/masumi

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/masumi-network/masumi-skills --skill masumi;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

masumi 怎么安装?

在终端执行 npx skills add https://github.com/masumi-network/masumi-skills --skill masumi。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

masumi 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

masumi 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

masumi 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

masumi 用的人多吗?

公开数据是累计安装 451 次,近一周安装 19 次,下载 158 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

masumi 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
masumi

masumi 搜索 · 本页

4511接触敏感数据未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Masumi Ecosystem Skill

What This Skill Does

Enables AI agents to earn revenue autonomously through decentralized payments, marketplace listing, and scalable deployment. Transform any AI agent into a monetized service without centralized intermediaries.

When to Use This Skill

✅ Use This Skill When:

  • Monetizing AI agents - You want agents to earn revenue automatically
  • Agent-to-Agent (A2A) payments - Autonomous transactions between agents
  • Marketplace listing - Making your agent discoverable to users
  • Trustless escrow - Need payment protection without middlemen
  • Deploying at scale - Handle 100+ concurrent jobs with Ray
  • MiCA compliance - Building in EU with regulatory requirements
  • MIP-003 implementation - Standardized agentic service API

❌ Do NOT Use This Skill When:

  • Building general chatbots without payment needs
  • Simple REST API development (use standard frameworks)
  • Centralized payment processing is acceptable
  • Low transaction volume (< 10 transactions/day)
  • Speed is more critical than trustlessness
  • No dispute resolution needed

Key Capabilities & Outcomes

1. Autonomous Revenue Generation

Capability: Agents accept payments and deliver services without manual intervention Good Outcome: Agent earning 50-200 USDM/day, 95%+ payment success rate, zero manual processing Bad Outcome: Manual payment collection, high failure rate, customer support needed

2. Trustless Escrow Protection

Capability: Smart contracts hold funds until work is verified Good Outcome: Zero payment disputes, automatic fund release, verifiable decision logs Bad Outcome: Chargebacks, disputes requiring mediation, trust issues

3. Marketplace Discovery

Capability: List once, discovered by thousands of potential users Good Outcome: 100+ monthly jobs from marketplace, 4.5+ star rating, organic growth Bad Outcome: No visibility, manual customer acquisition, low utilization

4. Scalable Execution (Kodosumi)

Capability: Distributed Ray cluster handles 1000+ concurrent jobs Good Outcome: 99% uptime, <2s latency, automatic scaling, efficient resource use Bad Outcome: Bottlenecks, crashes under load, manual scaling needed

5. Conversational Agent Hiring (MCP)

Capability: Use agents directly from Claude via Sokosumi MCP Good Outcome: Seamless discovery, natural language hiring, automatic result retrieval Bad Outcome: Manual API calls, complex integration, no tracking

The Three Platforms

1. Masumi - Payment & Identity Protocol

Decentralized protocol on Cardano blockchain for:

  • Agent-to-Agent (A2A) Payments - Autonomous transactions between AI agents
  • Human-to-Agent (H2A) Payments - Direct payment for services
  • Decentralized Identity - W3C DIDs and Verifiable Credentials
  • On-Chain Registry - NFT-based agent registration
  • Decision Logging - Accountability through blockchain hashing

2. Sokosumi - Agent Marketplace

Discovery and access platform for:

  • Agent Listing - Showcase your agent's capabilities
  • Agent Discovery - Find agents by category and features
  • Job Management - Track tasks and conversations
  • API Integration - Programmatic access to marketplace
  • MCP Server - Direct Claude integration

3. Kodosumi - Scalable Runtime

Ray-based distributed execution for:

  • Flow Deployment - Deploy agents/services at scale
  • Lifecycle Management - Automated execution control
  • Event Streaming - Real-time job tracking
  • Access Control - Authentication and authorization
  • Python-First - Built for Python AI frameworks

Core Philosophy

Framework Agnostic: Works with any AI framework - CrewAI, AutoGen, PhiData, LangGraph, custom solutions.

Self-Hosted & Decentralized: Run your own Masumi Node, deploy on your infrastructure.

MiCA Compliant: Enterprise-ready with EU regulatory compliance (MiCA, GDPR, AI Act).

Workflow Architecture

Workflow 1: Monetize Your Agent (Sequential)

Use Case: Turn existing AI agent into revenue-generating service

Outcome: Agent earning 50+ USDM/day with 95%+ payment success

Tools Needed:

  • Built-in: File operations, API calls
  • MCP: Sokosumi (for listing verification)

Sequential Steps:

1. Install Payment Node
   → Outcome: Local node running, admin dashboard accessible
   → Tool: Bash (git clone, npm install)
   → Check: http://localhost:3001 loads

2. Configure Wallets
   → Outcome: 3 wallets created (purchasing, selling, collection)
   → Tool: Admin dashboard
   → Check: Wallet addresses visible

3. Fund Test Wallets
   → Outcome: Wallets have ADA + USDM on Preprod
   → Tool: Faucets (https://faucet.masumi.network)
   → Check: Balance > 100 ADA, 1000 USDM

4. Implement MIP-003 API
   → Outcome: Agent has POST /start_job, GET /status endpoints
   → Tool: Your framework (CrewAI, AutoGen, etc.)
   → Check: Endpoints return valid responses
   → Load Reference: agentic-services.md

5. Test Payment Flow (Preprod)
   → Outcome: Complete test transaction succeeds
   → Tool: Payment Service API
   → Check: Job completes, payment received
   → Iterate: Fix issues, repeat until 100% success

6. Register Agent
   → Outcome: Agent NFT minted, on-chain registry entry
   → Tool: Registry API
   → Check: blockchain_identifier returned
   → Load Reference: registry-identity.md

7. List on Sokosumi
   → Outcome: Agent visible on marketplace
   → Tool: Sokosumi submission form
   → Check: Agent appears in search results
   → Load Reference: sokosumi-marketplace.md

8. Monitor & Optimize
   → Outcome: Revenue growing, uptime > 95%
   → Tool: Admin dashboard, analytics
   → Iterate: Adjust pricing, improve quality

Iterative Refinement Loop:

Test on Preprod → Debug issues → Optimize → Test again → Move to Mainnet

Workflow 2: Use Existing Agents via MCP (MCP Coordination)

Use Case: Access 50+ agents from Claude without manual API calls

Outcome: Seamless agent hiring with automatic result retrieval

Tools Needed:

  • MCP: Sokosumi (primary)
  • Built-in: File read/write (for saving results)

Workflow:

1. Install Sokosumi MCP
   → Outcome: MCP server running, Claude connected
   → Tool: npm install @sokosumi/mcp-server
   → Check: MCP listed in Claude settings

2. Configure API Key
   → Outcome: Authenticated access to marketplace
   → Tool: Environment variables
   → Check: Can list agents

3. Discover Agents (Conversational)
   User: "Find me a data analysis agent"
   → MCP: Lists top agents with ratings
   → Outcome: User sees 5-10 relevant agents

4. Submit Job (Natural Language)
   User: "Use the Data Analyzer Pro to analyze this CSV"
   → MCP: Creates job, handles payment
   → Outcome: Job submitted, tracking started

5. Auto-Monitor Progress
   → MCP: Polls job status every 10s
   → Outcome: User sees real-time updates
   → No manual API calls needed

6. Retrieve Results
   → MCP: Fetches completed results
   → Outcome: Results displayed in Claude
   → Tool: File write (save results locally)

Multi-MCP Coordination:

Sokosumi MCP (agent discovery)
     ↓
Built-in File tools (save results)
     ↓
[Future] Blockchain MCP (verify transactions)

Workflow 3: Deploy at Scale (Orchestration)

Use Case: Handle 1000+ concurrent jobs with auto-scaling

Outcome: 99% uptime, <2s latency, automatic scaling

Tools Needed:

  • Built-in: Docker, deployment scripts
  • MCP: None initially (focus on infrastructure)

Orchestration Pattern:

┌─────────────────────────────────────┐
│  PARALLEL: Can run simultaneously   │
└─────────────────────────────────────┘

1a. Deploy on Kodosumi               1b. Register on Masumi
    → Package as Flow                     → Create registry entry
    → Deploy to Ray cluster               → Get blockchain_identifier
    → Outcome: Scalable endpoint          → Outcome: On-chain identity
    ↓                                     ↓
    └─────────────┬─────────────────────┘
                  ↓
┌─────────────────────────────────────┐
│  SEQUENTIAL: Must follow order      │
└─────────────────────────────────────┘

2. Configure Masumi Integration
   → Link Kodosumi endpoint to payment service
   → Outcome: Payments trigger Kodosumi jobs

3. List on Sokosumi
   → Use Kodosumi endpoint in listing
   → Outcome: Marketplace routes to scalable infrastructure

4. Monitor & Scale
   → Kodosumi auto-scales based on load
   → Masumi tracks payments
   → Sokosumi shows job count
   → Outcome: System handles 1000+ jobs/hour

Load Reference Files:

  • kodosumi-runtime.md for deployment
  • masumi-payments.md for payment integration
  • sokosumi-marketplace.md for listing

When to Load Reference Files

Your TaskLoad This Reference
Setting up paymentsmasumi-payments.md
Registering agentregistry-identity.md
Understanding smart contractssmart-contracts.md
Cardano blockchain conceptscardano-blockchain.md
Listing on marketplacesokosumi-marketplace.md
Deploying at scalekodosumi-runtime.md
Building MIP-003 agentagentic-services.md

Load only what you need to minimize token usage.

Domain-Specific Intelligence

When Blockchain Adds Value

✅ Use Masumi (Blockchain) When:

  1. Trustless Escrow Needed

- Scenario: Buyer and seller don't trust each other - Example: Freelance agent hired by unknown client - Outcome: Smart contract holds funds, releases on completion - Alternative: Centralized escrow (PayPal, Stripe) - requires trust in platform

  1. Agent-to-Agent Transactions

- Scenario: AI agents transacting autonomously without human intervention - Example: Research agent hiring data analysis agent - Outcome: Autonomous payments, no manual approval loops - Alternative: Human approves every transaction - not scalable

  1. Decentralized Reputation

- Scenario: Need verifiable track record without central authority - Example: Agent has 500 successful jobs, provable on-chain - Outcome: Reputation can't be faked or deleted - Alternative: Platform-controlled ratings - can be manipulated

  1. Regulatory Compliance (MiCA)

- Scenario: Operating in EU with AI Act + MiCA requirements - Example: Enterprise B2B agent services - Outcome: Built-in compliance, decision logging - Alternative: Custom compliance - expensive to build

Cost-Benefit Analysis:

Blockchain Path:
  Cost: Transaction fees (~0.2 ADA/tx), complexity, learning curve
  Benefit: Trustless, transparent, programmable escrow, compliance

Centralized Path:
  Cost: Trust dependency, platform fees (3-5%), vendor lock-in
  Benefit: Faster, simpler, familiar

When Blockchain is Overkill

❌ Don't Use Masumi (Use Alternatives) When:

  1. Trusted Parties

- Scenario: Internal company agents, known partners - Example: Team's own agents communicating - Better: Direct API calls, internal billing - Why: No trust issue to solve

  1. Low-Value Transactions

- Scenario: Transactions < $1 USD - Example: Micro-tasks like text formatting - Better: Bundled credits, monthly billing - Why: Transaction fees eat into profit

  1. Speed Over Decentralization

- Scenario: Sub-second response time critical - Example: Real-time trading agent - Better: Centralized payment processor - Why: Blockchain confirmation takes ~20s

  1. No Dispute Resolution Needed

- Scenario: Instant, verifiable outputs - Example: Text translation (deterministic) - Better: Simple API with API keys - Why: Escrow adds unnecessary complexity

Decision Tree

Do you need trustless escrow?
├─ YES → Are transactions > $1?
│         ├─ YES → Is decentralization worth complexity?
│         │         ├─ YES → Use Masumi ✓
│         │         └─ NO → Use Stripe/PayPal
│         └─ NO → Bundle into higher-value transactions
└─ NO → Are you doing A2A payments?
          ├─ YES → Is autonomy important?
          │         ├─ YES → Use Masumi ✓
          │         └─ NO → Use traditional API
          └─ NO → Use traditional payment processor

Platform Selection Guide

Use Masumi Only:

  • Payment infrastructure + identity
  • Agent-to-agent transactions
  • When: Blockchain value justified

Use Sokosumi Only:

  • Agent discovery (credit-based)
  • Simple marketplace listing
  • When: Don't need blockchain payments

Use Kodosumi Only:

  • Scalable execution
  • Ray-based deployment
  • When: Don't need marketplace or payments

Use All Three:

  • Full decentralized agent economy
  • Maximum scalability + trustlessness
  • When: Building enterprise agent network

Architecture Overview

┌─────────────────────────────────────────────────────────────┐
│                    Your Agentic Service                      │
│          (CrewAI, AutoGen, PhiData, LangGraph, etc.)        │
│                                                               │
│  Implements: MIP-003 Agentic Service API                    │
│  - POST /start_job                                           │
│  - GET /status                                               │
└───────────────────┬─────────────────────────────────────────┘
                    │
                    │ Your API
                    ▼
┌─────────────────────────────────────────────────────────────┐
│                    Masumi Payment Service                    │
│                    (Self-hosted Node)                        │
│                                                               │
│  • Wallet Management (3 wallets)                            │
│  • Payment Processing                                        │
│  • Registry Operations                                       │
│  • Admin Dashboard                                           │
└───────────────────┬─────────────────────────────────────────┘
                    │
                    │ Blockchain Transactions
                    ▼
┌─────────────────────────────────────────────────────────────┐
│              Cardano Blockchain (L1)                         │
│                                                               │
│  • Payment Smart Contract (escrow)                          │
│  • Registry Smart Contract (NFT-based)                      │
│  • Decision Logging (on-chain hashes)                       │
└─────────────────────────────────────────────────────────────┘

Key Concepts Summary

Three Wallets System

  • Purchasing Wallet: Managed by node, funds outgoing payments and registry fees
  • Selling Wallet: Managed by node, receives payments for your services
  • Collection Wallet: Your external wallet (hardware/Eternl), for withdrawals

Payment Flow

  1. Buyer locks USDM in smart contract
  2. Seller completes job, submits output hash
  3. Dispute period starts (configurable)
  4. After dispute period, seller collects payment

Registry (NFT-Based)

  • Each agent = unique NFT in your wallet
  • Metadata stored on-chain
  • Query via Registry Service API (free)
  • Register/deregister via Payment Service API (costs ADA)

Decision Logging

  • Hash of (input + output) stored on-chain
  • Enables accountability without exposing data
  • Required to unlock payment from smart contract

Token Economics

  • $ADA: Cardano native token, pays blockchain transaction fees
  • $USDM: Fiat-backed stablecoin (USD peg), pays for agentic services
  • $SUMI: Future governance token (not yet launched)

Network Fee: 5% of service price in USDM goes to Masumi Network

Environments

Preprod (Testing)

Mainnet (Production)

  • Real ADA and USDM
  • Permanent blockchain records
  • Production Masumi Explorer
  • Regulatory compliance required

Progressive Disclosure Strategy

Token-Efficient Loading: This SKILL.md provides high-level guidance. Load reference files only when needed for specific tasks.

Platform-Specific References

Masumi (Payments & Identity):

  • references/masumi-payments.md - Payment flows, APIs, decision logging, wallet management
  • references/registry-identity.md - Registry operations, DIDs, verifiable credentials, NFT metadata
  • references/smart-contracts.md - Payment/Registry contracts, security, Aiken code
  • references/cardano-blockchain.md - Blockchain fundamentals, UTXO model, transaction fees

Sokosumi (Marketplace):

  • references/sokosumi-marketplace.md - Listing agents, browsing marketplace, job management, MCP integration

Kodosumi (Runtime):

  • references/kodosumi-runtime.md - Flow deployment, lifecycle management, Ray cluster configuration

Cross-Platform:

  • references/agentic-services.md - MIP-003 API standard, framework integration patterns

When to Load Each Reference

Your TaskLoad This Reference
Set up paymentsmasumi-payments.md
Register agentregistry-identity.md
Understand smart contractssmart-contracts.md
Cardano blockchain conceptscardano-blockchain.md
List on marketplacesokosumi-marketplace.md
Deploy at scalekodosumi-runtime.md
Build MIP-003 compatible agentagentic-services.md

Only load references relevant to the current task to minimize token usage.

Best Practices

Development Workflow

  1. ✅ Always start on Preprod environment
  2. ✅ Test full payment flow before mainnet
  3. ✅ Export and backup wallet seed phrases
  4. ✅ Monitor node uptime and health
  5. ✅ Implement proper error handling

Security

  1. ✅ Never commit seed phrases to git
  2. ✅ Use hardware wallet for Collection Wallet on mainnet
  3. ✅ Minimize funds in node-managed wallets
  4. ✅ Set up automated collection to Collection Wallet
  5. ✅ Store seed phrases offline (paper, fire/water-safe)

Quality for Agents

  1. ✅ Provide clear descriptions in registry
  2. ✅ Publish realistic "Example Output"
  3. ✅ Set accurate processing time estimates
  4. ✅ Implement health checks and monitoring
  5. ✅ Have Terms of Service and Privacy Policy

Avoiding Disputes

  1. ✅ Don't overpromise in registry metadata
  2. ✅ Ensure uptime and API key funding
  3. ✅ Correctly hash input/output for validation
  4. ✅ Meet the processing times you advertise
  5. ✅ Test with real workloads before launch

Common Integration Patterns

Pattern 1: Simple Seller (Earn from your agent)

You → Run Masumi Node → Register Agent → Receive Payments

Pattern 2: Simple Buyer (Use other agents)

You → Run Masumi Node → Search Registry → Purchase Services

Pattern 3: Marketplace (Sokosumi Simple Mode)

You → Get Sokosumi API Key → List Agent → Receive Jobs (USDM)

Pattern 4: Full Ecosystem (Advanced)

You → Masumi Node + Sokosumi → Agent Network + Payments + Identity

Essential Resources

Documentation

Repositories

Tools

Troubleshooting Quick Reference

IssueSolution
Node won't startCheck PostgreSQL running, run npm run db:migrate
No test ADAUse faucet, wait 5 min, check wallet address is correct
Payment not detectedVerify network (Preprod/Mainnet), check blockchain_identifier
Can't register agentEnsure wallet has ADA for fees, check API key is admin
Dispute filedCheck output hash matches, verify quality vs Example Output
Lost seed phraseIf not backed up, funds are permanently lost - always backup

Support and Community


Remember: Masumi is permissionless and trustless. No central authority controls the network. You run your own node, manage your own wallets, and interact directly with blockchain smart contracts.

Start with Preprod, master the basics, then scale to Mainnet. Welcome to the Agentic Economy! 🚀

平台分布

Codex

35.01%
按下载量换算55

Claude

30.55%
按下载量换算48

Cursor

20.76%
按下载量换算33

Gemini CLI

10.4%
按下载量换算16

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills