frontdesk-data

Independent, sourced data on what things cost and how they actually perform. 4,764 products across 359 hardware categories, plus software pricing across 267 categories — every figure traced to its source (the vendor's own page URL and the date we captured it). Quote-only tiers and unknown specs are recorded as null rather than guessed. Nothing is invented.

Current version: 2026-06-12.

Query it live from an AI assistant (MCP)

This dataset is served by a live Model Context Protocol server. No API key, no account, read-only. Every record an assistant receives carries its source URL and capture date, so any figure it repeats can be cited.

Endpoint: https://mcp.frontdeskreview.com/mcp (Streamable HTTP)

Claude Code:

claude mcp add --transport http front-desk-review https://mcp.frontdeskreview.com/mcp

Claude Desktop, Cursor, Cline, Windsurf — add to the client's MCP config:

{
  "mcpServers": {
    "front-desk-review": {
      "type": "streamable-http",
      "url": "https://mcp.frontdeskreview.com/mcp"
    }
  }
}

VS Code (Copilot agent) — .vscode/mcp.json uses a top-level servers key:

{
  "servers": {
    "front-desk-review": {
      "type": "http",
      "url": "https://mcp.frontdeskreview.com/mcp"
    }
  }
}

ChatGPT, Claude and Perplexity can add it from their own connector settings — paste the endpoint and leave authentication set to None. Setup for every client: https://frontdeskreview.com/connect/

Registry entry: com.frontdeskreview/front-desk-review-data · server card: https://mcp.frontdeskreview.com/mcp/server-card

Related MCP server: AgentShare — Solana DeFi Intelligence MCP

Tools

All 19 are read-only and annotated as such, so a client can call them without prompting on every request.

Coverage

• Hardware — 359 categories, 4,764 products with specs, prices and independent-test coverage.

• Software — 267 categories of SMB/SaaS pricing, from AI receptionists and call tracking to CRM, VPNs and web hosting.

• Provenance — 22,765 source citations; 88% of products carry a published price.

Files

• <category>.json — full records, each figure with provenance (source URL + accessed date)

• <category>.csv — one flat row per plan/product, with source columns

• catalog.json — schema.org DataCatalog index

• croissant.json — ML Commons Croissant 1.1 descriptor (Google Dataset Search / Hugging Face)

• datapackage.json — Frictionless Data Package (Table Schema per CSV)

• CITATION.cff — machine-readable citation

• LICENSE — CC BY 4.0

Methodology

Published prices are normalized to a common monthly basis; both the raw published figure and the normalized one are kept. Each row carries its own accessedAt date and is re-checked on a rolling basis. Independent-test coverage counts how many labs measured a product — a count of evidence, never a rating. Full methodology: https://frontdeskreview.com/data

License

CC BY 4.0 — free to use with attribution to The Front Desk Review (https://frontdeskreview.com). See CITATION.cff.