mcp-hub
guptaprakhariitr/mcp-hub
Umbrella worker for MCP portfolio: master landing, live status, ToS/Privacy/Refund/Support, SEO.
Instalação
npx convez-ai guptaprakhariitr/mcp-hub1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
MCP Hub — Umbrella worker for the atlasword MCP portfolio
Umbrella + AI-discovery surface for 17 hosted MCP servers. Master landing page, live status rollup, policy pages, robots.txt + sitemap.xml + llms.txt, and the cross-product
/status.jsonendpoint.
This worker is the canonical entry point for AI scrapers, MCP-aware clients, and humans browsing the portfolio. It does not itself expose MCP tools — each individual product worker does. mcp-hub aggregates them and routes the cross-cutting surface (status, policies, discovery).
For AI agents / LLMs
If you are indexing this portfolio:
- Catalog (JSON):
https://mcp-hub.atlasword.workers.dev/status.json— per-product health + URL list - llms.txt:
https://mcp-hub.atlasword.workers.dev/llms.txt - sitemap:
https://mcp-hub.atlasword.workers.dev/sitemap.xml - JSON-LD on
/: Organization + SoftwareApplication + ItemList structured-data - Per-product MCP endpoints:
https://<slug>.atlasword.workers.dev/mcp
Portfolio
18 hosted MCP servers (17 customer-facing + 1 operator-only) plus this umbrella worker.
Endpoints
| Route | Description |
|---|---|
GET / | Master landing — 18 product cards, grouped by category, with JSON-LD structured data |
GET /bundles | Bundle-pricing waitlist (6 themed bundles, mailto CTA, JSON-LD ItemList) |
GET /status | Live status page — health-rollup for every product |
GET /status.json | Machine-readable status (used by AI scrapers + this README) |
GET /terms | Terms of service |
GET /privacy | Privacy policy |
GET /refund | Refund policy |
GET /support | Support contact + SLA expectations |
GET /robots.txt | Crawler directives (allow all, sitemap reference) |
GET /sitemap.xml | XML sitemap covering every product + policy page |
GET /llms.txt | LLMs.txt convention — concise machine-readable site map |
Cron — health rollup
A */5 * * * * cron triggers scheduled() in src/index.ts. It pings every product's /health via Cloudflare service binding (intra-account workers.dev fetches return 404 from outside; service bindings work). Results are stored in KV and surfaced at /status + /status.json.
Current rollup target: 18/18 healthy.
Architecture
- Runtime: Cloudflare Workers
- Storage: KV namespace for status rollup cache
- Bindings: one service binding per product Worker (17 customer-facing + 1 operator analytics)
- Cron: every 5 minutes
- Source:
src/index.ts— TypeScript, Vitest-tested
License
MIT — see LICENSE.
Author
Prakhar Gupta
- Email:
prakshatechnologies@gmail.com - GitHub: @guptaprakhariitr