Skip to content
The Direct Index logo

The Direct Index MCP server

Which MCP servers are The Direct Index's own, from the official registry and from asking its own domain, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.

read 2026-10-27

← Back to The Direct Index

Whose server it is
  • Registry nameorg.directindex/scannerverified in the MCP registry

The registry name sits under directindex.org. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

When we asked itchecked 2026-10-27
  • Stateanswerschecked 2026-10-27
  • Tools20
  • Server calls itselfdirect-index
  • Server version0.2.0
  • Protocol version2025-06-18

On 2026-10-27 the server answered initialize and tools/list without a sign-in. First answer in 54 ms.

Should an agent connect itNot proven either way yet
  • Whether its tools do the product's job has not been judged yet.
  • Checked on 2 of the 7 days it takes to judge reliability.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

Measured dailydaily checks: 2, the first on 2026-10-27
  • INITE MCP scoreafter a weeka score needs 7 daily checks: 2 so far
  • Answeredmeasuring2 of the 7 daily checks it takes to print a share
  • Median first answer282 ms
  • Space in an agent's context≈ 10,849 tokensthe definitions of all its tools (20), which an agent carries on every turn
  • Tool definitionsA · 97 of 100
  • Typed results0 of 20 toolsdeclared outputSchema: results come typed, read once
  • Parameters without a description: in 4 of 20

Worth a look

None of the five warning signs in these definitions.

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

Tools it offers20 tools · read 2026-10-27
  • scan_website

    Scan a business website and return its Agent-Direct Score (0 to 100), grade, findings and fixes. Reuses a scan under 24 hours old; otherwise runs one, which takes 10 to 40 seconds.

  • get_site_report

    Return the latest Agent-Direct report for a domain that has already been scanned, including findings, fixes and agent-discoverability checks.

  • city_leaderboard

    Businesses in a city index ranked by Agent-Direct Score, with grade, menu format, direct channel and whether AI crawlers are blocked.

  • find_direct_places

    Places in a city that take a booking or an order on their own website, best Agent-Direct Scores first, with the direct booking or ordering link for each (the owner's corrected link where the business is claimed), the vendor behind it…

  • menu_prices

    What things cost on restaurant menus in a city: the middle price, and the middle half of prices, of an espresso, a cappuccino, a beer, a glass of wine, a margherita, a burger, a main course, a lunch offer, a cola and a bottle of still…

  • list_cities

    List the city indexes that exist, with counts and status.

  • order_deep_report

    Order the paid deep report (5 EUR, one-off) for a business in a city index: ChatGPT, Claude and Gemini are asked ten questions about it the way people ask, every answer is kept in full with where it sent people, plus the nearest similar…

  • get_deep_report

    Status of a deep report order by its token, and the finished report itself as Markdown once it is done (every answer in full, the table of where answers sent people, the nearest similar places, the fix plan and the kit).

  • what_assistants_say

    What ChatGPT, Claude and Gemini answer about one business: the latest verdict per question (official website, where to book, where to order: own site, vendor, marketplace, directory, no link), the answers quoted, the questions people ask…

  • assistant_recommendations

    The places ChatGPT, Claude and Gemini name in a city when asked the way people ask ('best pizza delivered right now', 'book me a table for Saturday'): per question, the places named, which assistant named each, whether with a link to the…

  • assistant_destinations

    Where the assistants send people in a city: the league table of destinations (the businesses' own sites, each marketplace and directory by name, press and guides, other sites) with the share of answers that link each one, per assistant…

  • compare_cities

    Two city indexes side by side: business websites scanned, the share that take a booking or an order on their own site, the share that turn an AI crawler away, the share scoring A or B, grades, menus, channels, builders, vendors, the…

  • country_summary

    A country's indexes added together (ISO code, for example PT or DE): sites scanned, direct share, crawler-blocked share, A-or-B share, the cities with their numbers, the kinds, the vendors common in the country. Without a code: every…

  • rankings

    Every city index of a kind ranked on three measures with its sample size and date: direct share (a booking or ordering path on the business's own domain), crawler-open share (not turned away at the edge) and A-or-B share. Indexes with…

  • search_places

    Find businesses in the city indexes by name or domain, filtered by city, kind, grade, builder, booking or ordering vendor and finding (for example menu_pdf, edge_blocks, no_structured, booking_marketplace). Fifty results a page, best…

  • direct_record

    The direct record of a business: the booking, ordering and menu links an agent should use instead of a marketplace, where each leads (the business's own site or a named vendor), whether and when a browser verified it, the hours…

  • vendor_report

    A booking or ordering vendor across the index: software on the vendor's domain or a marketplace, booking or ordering, how many sites use it, their average score and A or B share, how the browser action test went on them, the cities…

  • city_facts

    What a city index's websites say, compared with every index of the same kind: the share of listed businesses with no web address, only a social profile or their own site; the share of sites that turn an AI crawler away, publish a readable…

  • city_changes

    What changed in a city index over the last days: findings fixed or new, grades up or down, AI crawlers turned away or let in, direct booking or ordering channels and vendors gained or lost, and assistants that changed their answer about a…

  • vendor_terms

    Whether a booking or ordering vendor's terms let an AI agent book or order through it: allowed, partners only, banned, or not stated; with the vendor's own page and the date the terms were read. Without a vendor: every researched vendor…

How to connect

Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:

{
  "mcpServers": {
    "scanner": {
      "type": "http",
      "url": "https://directindex.org/mcp"
    }
  }
}

Add it

Claude Code, in a terminal

claude mcp add --transport http directindex https://directindex.org/mcp

Add to CursorAdd to VS Code

claude.ai, ChatGPT and other clients: add https://directindex.org/mcp as a custom connector in the settings.

OpenAPI spec

33 paths, read 2026-10-27. Open the spec ↗

Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.