For AI agents and power users

Use our reviews from your AI assistant

Everything on a review page is also available to software: the verdict, rating, evidence level, dated prices, scores, and sources. An agent gets exactly the facts you see, with the same IDs, so its answers can link back to the line it quoted. No account or key is needed, and every tool is read-only.

Connect with MCP

Our Model Context Protocol server uses the Streamable HTTP transport at:

https://topratedsoftware.net/mcp

Claude Desktop and claude.ai

Settings → Connectors → Add custom connector, then paste the server URL.

https://topratedsoftware.net/mcp

Claude Code

claude mcp add --transport http topratedsoftware https://topratedsoftware.net/mcp

Cursor

Add to .cursor/mcp.json in your project, or ~/.cursor/mcp.json for every project.

{
  "mcpServers": {
    "topratedsoftware": {
      "url": "https://topratedsoftware.net/mcp"
    }
  }
}

VS Code

Add to .vscode/mcp.json.

{
  "servers": {
    "topratedsoftware": {
      "type": "http",
      "url": "https://topratedsoftware.net/mcp"
    }
  }
}

Clients that only support stdio

Bridge the remote server with mcp-remote.

{
  "mcpServers": {
    "topratedsoftware": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://topratedsoftware.net/mcp"
      ]
    }
  }
}

Tools

search_software: Find reviewed software by query, category, or franchise (VPN, antivirus…).
get_review: The full evidence record: verdict, scores, plans, sources, recent changes, citation, and fact IDs.
get_verified_pricing: Plans verified in the last 30 days plus price history.
get_recent_changes: Price increases and decreases, removed features, incidents, and acquisitions, with sources.
compare_software: Side-by-side facts for 2–6 products: the same table people see on /compare.
find_alternatives: Alternatives, flagged when the original product recently raised its price.
get_editorial_policy: Independence pledge, evidence standards, citation rules, and data license.

Try it without a client:

curl -s https://topratedsoftware.net/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"compare_software","arguments":{"ids":["nordvpn","surfshark"]}}}'

Fact IDs: the same fact on the page and in the data

Each evidence record has a facts list. Every fact has an id such as verdict.rating, pricing.verifiedAt, or sources.2, and a url that jumps to that fact on the review page, for example:

https://topratedsoftware.net/reviews/nordvpn#fact-verdict-rating

On the page the same element carries data-fact-id, so a person can check any number an assistant quotes.

Without MCP

llms.txt: an index of every in-depth review, written for language models.
llms-full.txt: the full text of every in-depth review in one Markdown file.
Any review as Markdown: add .md to its URL, e.g. /reviews/nordvpn.md.
/data/v1/catalog.json and /data/v1/reviews/<id>.json: static evidence records.
/api/v1: a JSON API with live verified pricing, change events, comparisons, and alternatives.

Limits and citation

Rate limits per IP address: 120 requests a minute to the MCP server, 60 a minute to the JSON API. Static files are not limited.
AI search engines, assistants, and agents may crawl, index, summarize, and cite our published reviews and data, provided answers link back to the source page.
Preferred citation format: NordVPN review — TopRatedSoftware (https://topratedsoftware.net/reviews/nordvpn)
Respect the evidence label. "Expert Researched" means we did not run hands-on tests, so please don't describe those reviews as tested.

Rankings can't be bought, through the API or anywhere else. Read our pledge.