skylight-mcp-remote

skylight-mcp-remote

Remote MCP deployment scaffold for Skylight's MCP server, hosted on Cloudflare Workers and Containers, enabling HTTP access to the stdio-based Ruby MCP server.

Category
访问服务器

README

skylight-mcp-remote

Remote MCP deployment scaffold for skylight-mcp, built on Cloudflare Workers plus Cloudflare Containers.

Relevant upstream links:

This repo is for people who want to expose the Ruby skylight-mcp server through a hosted HTTP endpoint instead of running it locally over stdio.

What It Does

  • Terminates HTTP requests at a Cloudflare Worker.
  • Protects /mcp with a shared bearer token.
  • Forwards authenticated traffic to a containerized Node bridge.
  • Runs gem exec skylight-mcp --token "$SKYLIGHT_MCP_TOKEN" inside the container.
  • Adapts hosted MCP traffic into the stdio-based Ruby server.

If you found this repo while searching for Skylight MCP, Skylight Ruby performance tooling, or the skylight-mcp gem, start with Skylight and the package pages on RubyGems or Ruby China.

MCP Behavior

The bridge is intentionally tolerant of real hosted-client behavior:

  • OPTIONS /mcp returns 204 for browser preflight requests.
  • HEAD /mcp returns 200 for authenticated reachability checks.
  • POST /mcp supports standard JSON-RPC requests such as initialize, tools/list, and tools/call.
  • notifications/initialized is accepted as a notification and returns 202.
  • If a client sends follow-up traffic without a local in-memory session, the bridge bootstraps a replacement session before forwarding the request.

Architecture

  • apps/worker: Cloudflare Worker request handling, auth, and CORS.
  • apps/bridge: HTTP-to-stdio bridge service that manages skylight-mcp child processes.
  • Dockerfile: minimal container image for the bridge.
  • wrangler.jsonc: Worker + container deployment config.

Request flow:

  1. Client sends POST /mcp with Authorization: Bearer <MCP_SHARED_BEARER_TOKEN>.
  2. Worker validates auth and forwards the request to the Cloudflare Container binding.
  3. Bridge creates or resumes a session, then proxies JSON-RPC over stdio to skylight-mcp.
  4. Bridge returns JSON-RPC responses, mcp-session-id headers when relevant, and browser-friendly CORS headers.

Prerequisites

  • Node 22+
  • npm
  • Docker with buildx
  • Ruby is only required if you want to run the bridge outside the container image
  • A Cloudflare account with Workers and Containers enabled
  • A valid Skylight MCP token

Configuration

Required secrets:

  • MCP_SHARED_BEARER_TOKEN
  • SKYLIGHT_MCP_TOKEN

Optional environment variables:

  • PORT default 8080
  • LOG_LEVEL default info
  • BRIDGE_REQUEST_TIMEOUT_MS default 30000

.env.example contains the expected variable names for local development.

Local Development

Install dependencies and run tests:

npm install
npm test
npm run build

Run the bridge directly:

export SKYLIGHT_MCP_TOKEN=your-skylight-token
export PORT=8080
npm run dev:bridge

Quick checks:

curl -i http://127.0.0.1:8080/healthz

curl -i \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"local-test","version":"1.0"}}}' \
  http://127.0.0.1:8080/mcp

Deploying To Cloudflare

  1. Authenticate Wrangler.
  2. Build the workspace artifacts copied into the bridge image:
npm run build
  1. Set Worker secrets:
npx wrangler secret put MCP_SHARED_BEARER_TOKEN
npx wrangler secret put SKYLIGHT_MCP_TOKEN
  1. Build and push the bridge image:
docker buildx build --platform linux/amd64 --load -t skylight-mcp-remote:amd64 .
npx wrangler containers push skylight-mcp-remote:amd64
  1. Update wrangler.jsonc with your pushed image digest.

  2. Review the account-specific values in wrangler.jsonc before deploy:

  • Worker name
  • container name
  • registry image reference under containers[].image
  1. Deploy:
npx wrangler deploy

Smoke Test

Set helpers:

export BASE_URL="https://<your-worker-subdomain>.workers.dev"
export MCP_SHARED_BEARER_TOKEN="<your-shared-bearer-token>"

Health check:

curl -i "$BASE_URL/healthz"

Initialize:

curl -i \
  -H "Authorization: Bearer $MCP_SHARED_BEARER_TOKEN" \
  -H 'content-type: application/json' \
  -H 'Mcp-Method: initialize' \
  -H 'MCP-Protocol-Version: 2025-11-25' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"smoke-test","version":"1.0"}}}' \
  "$BASE_URL/mcp"

Follow with notifications/initialized:

curl -i \
  -H "Authorization: Bearer $MCP_SHARED_BEARER_TOKEN" \
  -H 'content-type: application/json' \
  -H 'Mcp-Method: notifications/initialized' \
  -H 'MCP-Protocol-Version: 2025-11-25' \
  -H 'mcp-session-id: <session-id>' \
  -d '{"jsonrpc":"2.0","method":"notifications/initialized"}' \
  "$BASE_URL/mcp"

List tools:

curl -i \
  -H "Authorization: Bearer $MCP_SHARED_BEARER_TOKEN" \
  -H 'content-type: application/json' \
  -H 'Mcp-Method: tools/list' \
  -H 'MCP-Protocol-Version: 2025-11-25' \
  -H 'mcp-session-id: <session-id>' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' \
  "$BASE_URL/mcp"

Limitations

  • The repo is a deployment scaffold, not a published npm package.
  • wrangler.jsonc is intentionally operational and must be edited for your own Cloudflare account.
  • Session state is in-memory inside the bridge container. The Worker smooths over some hosted-client edge cases, but the architecture is still fundamentally a stateful bridge around a stdio server.

License

MIT. See LICENSE.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选