mcp-skills-registry
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-skills-registrylist all available skills from the registry"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
monorepo – MCPP Aggregation Server Example
Demonstrates the monorepo topology of MCPP 3.7: a single HTTP endpoint, with path routing to sub servers.
Shape
┌─────────────────────────────────────────────┐
客户端 │ monorepo(单一 HTTP server,端口 8787) │
│ │
/openspec/mcp ──► OpenSpec skills sub server │
/mattpocock/mcp ──► Matt Pocock skills sub server │
其他路径 ──► 404 │
└─────────────────────────────────────────────┘Each path is an independent MCP endpoint; skills, resources, and sessions are isolated from each other.
src/registry.tsis the single route registry shared by the local entry point, Cloudflare Worker, and tests.stdio has no URL/path concept, so it is not suitable for this aggregation form.
Related MCP server: MCP Skill Registry
Initialization
bun run initThis command installs dependencies, reproduces third-party skills according to skills.lock.json, and generates a static registry that can be bundled into the Worker.
Common Commands
bun dev # 构建 skills 后启动 http://127.0.0.1:8787/
bun run demo # 端到端 smoke 测试
bun run skills:sync # 按 lock 同步第三方 skills
bun run skills:update # 更新第三方 commit 并重新生成 registry
bun run deploy:dry-run # 验证 Cloudflare Worker bundle
bun run deploy # 部署 Cloudflare WorkerDirectory Structure
├── src/
│ ├── index.ts # 本地网关和纯 fetch handler
│ ├── registry.ts # 唯一 sub server 注册表
│ └── static-skills.ts # 静态 skills resource 挂载器
├── servers/
│ ├── openspec/
│ │ ├── server.ts
│ │ ├── skills/ # 生成、Git ignored
│ │ └── skills.generated.ts # 生成、Git ignored
│ └── mattpocock/
│ ├── server.ts
│ ├── skills/ # 生成、Git ignored
│ └── skills.generated.ts # 生成、Git ignored
├── install-skills.js # manifest + lock 同步器
├── generate-skills-registry.js # 静态 registry 生成器
├── skills.json # source 声明
├── skills.lock.json # 锁定 commit,提交进 Git
├── test/smoke.ts
├── worker.ts
└── wrangler.jsoncReproducible Skills Build
skills.json + skills.lock.json
│
▼
install-skills.js
│ copy 锁定 commit
▼
servers/<id>/skills/
│
▼
generate-skills-registry.js
│
▼
servers/<id>/skills.generated.ts
│ static import
▼
Bun / Worker bundleNeither the generated directory nor the generated registry are committed; a fresh clone fully restores them via bun run init. The sub server does not read node:fs at runtime, so the Worker bundle will include the SKILL.md content.
Cloudflare Deployment
bun run deploy:dry-run
bun run deployThe session registry is in isolated memory; for production with multiple concurrent instances, session state must be externalized to Durable Objects.
Related
Specification:
MCPP.mdSpecification package:
@peri-code/mcppHosting convention: agent-plugins.org
This server cannot be deployed
Maintenance
Related MCP Connectors
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
Cloudflare Workers MCP server: ai-model-router
Cloudflare Workers MCP server: claude-skill-validator
- UnifAPIOAuthcom.unifapi
Hosted MCP server for live public-data APIs and Skills for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceA universal MCP server that acts as a unified gateway for dynamically connecting and managing multiple MCP servers via a single HTTP endpoint.7 npm6MIT
- AlicenseNot gradedqualityBmaintenanceA self-hostable MCP server that turns a folder of skills into callable tools via MCP and REST APIs.2MIT
- FlicenseNot gradedqualityDmaintenanceAn MCP server that enables Claude to deploy full-stack web apps to Cloudflare, including databases, authentication, and file storage, directly through natural language.-
- AlicenseNot gradedqualityCmaintenanceAn MCP server that acts as an AI knowledge and skills hub, automatically discovering, indexing, and serving skill libraries to any MCP client with full-text search, rich metadata, and hot reload.MIT