Upshift
upshift-mcp
Servidor MCP de referencia de Upshift. Tres herramientas, desplegadas de forma remota, en la revisión 2026-07-28 de la especificación MCP.
Existe para hacer dos trabajos a la vez: ser genuinamente útil para un agente, y ser el ejemplo práctico que un comprador mira antes de pagarnos para construir el suyo. Cada precio que cotiza se genera desde la tienda en vivo, por lo que no puede inventar uno.
https://mcp.upshiftsites.com/mcp{
"mcpServers": {
"upshift": { "type": "http", "url": "https://mcp.upshiftsites.com/mcp" }
}
}O ejecuta el mismo servidor localmente, para clientes que lanzan un comando en lugar de llamar a una URL. Nada que clonar o compilar — está publicado en npm:
{
"mcpServers": {
"upshift": { "command": "npx", "args": ["-y", "upshift-mcp"] }
}
}El mismo binario desde un shell. stdio es el predeterminado; --http sirve el Worker completo en localhost en su lugar:
npx upshift-mcpHerramientas
Herramienta | Descripción |
| Obtiene una página en vivo e informa datos estructurados, llms.txt, robots.txt, presupuestos de título y meta, encabezados, cobertura de texto alternativo, HTTPS y peso del documento, con una puntuación de 0-100 y una solución para cada hallazgo. |
| Clasifica las 14 líneas de plantillas de sitios web según un oficio, con enlaces de demostración en vivo, enlaces de tienda y precios reales, y explica por qué cada una coincidió. |
| La tabla de precios real para trabajos de servidor MCP y para sitios web. |
Gratuito y limitado a 20 llamadas/minuto por IP.
Related MCP server: 20-tool MCP server
Diseño
src/
worker.ts Cloudflare entrypoint: routing, origin policy, rate limit
server.ts The three tools; the one place production layers are applied
audit.ts Audit rules — pure, so every rule is testable without a network
safe-fetch.ts The SSRF guard for the one tool that takes a caller's URL
match.ts Template matching
services.ts What we charge for MCP work
catalog.generated.ts Template lines + prices, generated from the store
landing.ts The page a human gets at /
evals/ 12 qa_pairs, run over the real protocol
test/ 32 tests: protocol conformance + the pure logicTrabajando en ello
npm install
npm run dev # http://127.0.0.1:8788/mcp
npm run ci:verify # typecheck + tests + evalswrangler dev necesita macOS 13.5+ y no arrancará en la máquina de compilación, por lo que npm run dev sirve el src/worker.ts de producción en Node contra un KV en memoria. Mismo manejador, mismas rutas, mismas herramientas — no una segunda implementación. El despliegue sigue siendo wrangler deploy.
Los precios vienen de ../upshift-agency-site/src/data/store.ts. Después de cambiarlos allí:
npm run gen:catalogDesplegando
En vivo desde 2026-08-16 en https://mcp.upshiftsites.com/mcp. El espacio de nombres KV y el dominio personalizado están aprovisionados; npm run deploy envía una nueva versión.
Listado
Registro oficial de MCP:
com.upshiftsites/mcp, publicado comoactivecontra el espacio de nombres verificado por DNS.Glama — aprobado el 2026-08-16.
npm:
upshift-mcp—npx upshift-mcpejecuta este mismo servidor localmente.
Brechas honestas
En docs/VALIDATION.md — qué está verificado, qué no, y los cuatro errores reales que la suite de evaluación detectó durante la compilación (uno de ellos una omisión de seguridad). Ese documento es el argumento para enviar evaluaciones junto con un servidor, que es la parte de esto que vendemos.
Licencia
MIT. Toma cualquier cosa útil. Si prefieres que lo construyamos y ejecutemos por ti, eso está en https://upshiftsites.com/store/mcp-server-service/.
Available Tools
3 toolsupshift_quotePrice a jobARead-onlyInspect
Return Upshift's real prices. job_type 'mcp' covers MCP server work (registry listing, live tool preview, spec upgrade, full build); 'website' covers website templates and done-for-you launches. These are the actual listed prices, not estimates. Use when someone asks what a job costs, what a tier includes, or how long it takes. The figures are generated from the live store catalog, so this cannot quote a price that is not sold; it makes no network call and the same arguments always return the same answer.
Inputs: job_type (required), scope. Returns: note, offers, jobType
upshift_site_auditAudit a website for machine legibilityARead-onlyInspect
Fetch a live website and report how legible it is to search engines and AI assistants: structured data (JSON-LD), llms.txt, robots.txt, title and meta description, headings, image alt text, HTTPS and document weight. Returns findings with severity and a 0-100 score. Use this to check a real site before recommending changes to it. Makes one outbound request to the URL given, so it only sees what a fetch can establish — never rendering or browser metrics. Refuses private, loopback and link-local addresses, and reports a page as unreachable rather than scoring the error body.
Inputs: url (required). Returns: url, score, summary, findings, +2 more
upshift_template_matchFind the website template for a tradeARead-onlyInspect
Given an industry or trade (and optionally what the business needs), rank Upshift's 14 website template lines and return the best three, each with a live demo link, the store link, real prices, and the words that made it match. Use when someone asks what a website for a given trade should look like or cost. Answers from a catalog held in the server, so it makes no network call and returns no matches — rather than a forced one — when the trade is outside the 14 lines.
Inputs: industry (required), needs. Returns: query, matches
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 3 tools
Each tool targets a completely distinct purpose: site audit checks existing websites, template match recommends templates by trade, and quote provides pricing. There is no overlap in functionality or output.
All tools use the 'upshift_' prefix and then a clear noun: 'site_audit', 'template_match', 'quote'. The pattern is consistent, though 'quote' is a single word while the others are verb_noun compound.
Three tools is an ideal scope for a narrow commercial service. Each tool is necessary for the core workflow (audit, recommend, quote), and there is no room for bloat.
The set covers the essential sales workflow: audit existing site, recommend a template, and provide pricing. A minor gap is lack of a tool to get detailed information about a specific template, but the match tool returns demo links and prices.
Maintenance
Related MCP Connectors
Robots-aware website audits, pre-crawl quotes, and structured evidence for people and agents.
Scans any public website for SEO, AI search, trust, conversion and speed, and lists our prices.
Live 62-rule SEO, AEO & GEO audit, RFC 9727 api-catalog, llms.txt & AI citation tools.
19 free website QA and AI-visibility tools. Remote HTTP MCP, no account, no API key.
Related MCP Servers
AlicenseNot gradedqualityDmaintenanceChecks whether a website is readable and citable by AI search engines — llms.txt, Schema.org structured data, AI-bot access in robots.txt, content freshness, answer directness, E-E-A-T signals, plus a LocalBusiness Rich Results validator. Free, no API key, remote Streamable HTTP.1MIT- AlicenseNot gradedqualityDmaintenance20-tool MCP server - production web-intelligence API gateway exposing 10 Website Intelligence + 7 Marketing APIs as 20 MCP tools over Streamable HTTP + stdio, with pay-per-call Solana USDC payments. 20 tools: URL-to-Markdown, metadata, technology detection, contact extraction, SEO audit, SSL, DNS, sitemaps, robots.txt, Google Maps reviews, AI company/website summaries, citation checking Cloud-hostMIT

@operstack/mcpofficial
AlicenseAqualityBmaintenanceEnables Claude, Cursor, and other MCP clients to audit public sites across six areas, check and follow llms.txt links, compare rival sites side by side, and run sixteen content gates on local Markdown or MDX using only free public signals.438 npmMIT
io.github.spalofficial
AlicenseNot gradedqualityCmaintenanceEnables AI agents and CI systems to run deterministic website QA audits over MCP, with every finding backed by evidence and every result hashed and signed. It checks HTTP, TLS, security headers, SEO, links, accessibility, and robots.txt, with configurable pass/fail verdicts.MIT