Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计异常

conceptsconcepts 搜索

Agent Skill

concepts 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。

总安装

768

周安装

32

GitHub Stars

2

下载量

256
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:concepts(concepts 搜索)
来源仓库:https://github.com/contextvm/cvmi
仓库路径:skills/concepts
安装命令:
npx skills add https://github.com/contextvm/cvmi --skill concepts
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/contextvm/cvmi --skill concepts

简介

concepts 提供 ContextVM 协议的核心概念与常见问题解答。

  • 适合理解其去中心化通信机制及在 Nostr 上运行 MCP 服务的原理时使用。
  • 涵盖传输层抽象、密钥管理与消息路由等关键设计要点说明。
  • 建议配合其他 cvmi 技能共同使用以构建完整认知框架。
  • concepts 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

ContextVM Concepts

Core concepts, architectural decisions, and common questions about ContextVM.

What is ContextVM?

ContextVM is a protocol that liberates the Model Context Protocol (MCP) by running it over Nostr—a simple, open communication network built on cryptographic, censorship-resistant, and permissionless foundations.

Rather than relying on centralized infrastructure like domains, OAuth, or cloud hosting, ContextVM allows anyone to run or access services using only Nostr and an internet-connected device.

Key Insight: ContextVM operates at the transport layer, meaning existing MCP servers and clients can be used without code changes through the Gateway and Proxy tools.

Why Run MCP Over Nostr?

Running MCP over Nostr eliminates traditional infrastructure barriers:

Traditional RequirementNostr Solution
Domain name + DNSNot needed
Static IP addressNot needed
OAuth/API keysBuilt-in public key cryptography
Public hostingAny device with internet
Port forwardingOutbound-only relay connections

Nostr provides:

  • Identity via public/private key cryptography
  • Discovery through service announcements on relays
  • Transport via signed and encrypted events
  • Payments (optional) using Bitcoin and Lightning Network

Decentralization Model

ContextVM uses Nostr relays as a distributed message bus:

Client ⇄ Nostr Relay(s) ⇄ Server
  • No central directory or gatekeeper
  • Anyone can run a server and announce it (or keep it private)
  • Servers can connect to multiple relays for redundancy
  • Services can go offline and come back online without breaking references

The Dual API Advantage

ContextVM presents a unique dual API that lets you write your server once and make it accessible to both humans and machines:

For Developers: Build web apps, desktop applications, or CLI tools that interact with ContextVM servers directly using standard MCP patterns.

For AI Agents: LLMs operate the same service naturally through MCP's self-documenting capabilities (tools/list, schemas, etc.).

Build Once, Deploy Everywhere: Your service becomes a reusable component accessible through code, web interfaces, or AI agents.

Public vs Private Servers

FeaturePublic ServerPrivate Server
AnnouncementsPublished to relaysNot published
DiscoveryVia relay queriesKnown public key only
Access ControlOpen or whitelistedWhitelisted
EncryptionE2E encryptedE2E encrypted
PaymentsOptionalOptional

Use Private Servers For: Personal tools, team infrastructure, sensitive operations, development/testing.

Network Topology

Communication flows through three actors:

  1. Client: MCP client using ContextVM Proxy or SDK
  2. Relay(s): WebSocket servers routing encrypted events
  3. Server: Service using ContextVM Gateway or SDK

Flow:

  1. Server optionally publishes announcement to relays
  2. Client discovers service (public) or uses known pubkey (private)
  3. Client sends encrypted request via relays
  4. Server receives, decrypts, processes, responds

All messages are signed and end-to-end encrypted (NIP-44).

ContextVM vs Traditional Remote MCP

RequirementTraditional MCP (HTTP/SSE)ContextVM
Domain nameRequiredNot needed
DNS configurationRequiredNot needed
Static IPRequiredNot needed
Port forwardingRequiredNot needed
TLS certificateRequiredImplicit via encryption
AuthenticationOAuth/API keysBuilt-in pubkey crypto
HostingCloud VM/VPSAny device
DiscoveryCentralized directoriesDecentralized relays
Censorship resistanceLowHigh

Authentication

Authentication is built into Nostr's public key cryptography:

  • Every request is signed by the client's private key
  • Servers verify signatures to confirm identity
  • Server operators can:

- Allow all signed requests (open access) - Whitelist specific public keys (private access) - Require payment before processing

No OAuth, passwords, or API keys needed.

Security

All client-server communication is end-to-end encrypted using NIP-44:

  • Messages encrypted to recipient's public key
  • Only intended party can decrypt
  • Compromised relays cannot read content or impersonate parties
  • Ensures confidentiality, integrity, and non-repudiation

MCP Possibilities Beyond AI

MCP is a protocol for invoking remote functions—any computational task, not just AI:

  • SSH Access Portal: Secure remote machine access
  • Encryption as a Service: GPG operations in sandboxed environment
  • Data Processing: Validate, transform, analyze data on demand
  • Code Sandbox: Execute untrusted code safely
  • IoT Command Hub: Trigger physical actions remotely
  • Math & Simulation: Complex calculations or symbolic math

Getting Started

Deploy a Server (Gateway)

gateway-cli --private-key "your-key" \
  --relays "wss://relay.nostr.org" \
  --server "python my-mcp-server.py" \
  --public  # omit for private server

Connect a Client (Proxy)

proxy-cli --private-key "your-key" \
  --relays "wss://relay.nostr.org" \
  --server-pubkey "npub1..."

Build Native (SDK)

Use @contextvm/sdk for TypeScript applications with NostrServerTransport and NostrClientTransport.

References

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude

33.68%
按下载量换算86

Codex

33.14%
按下载量换算85

Cursor

18.57%
按下载量换算48

Gemini CLI

10.12%
按下载量换算26

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

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 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills