sofiaristoranteroma.it is partly visible to agents.
Some AI assistants are turned away by the website of Sofia. It can read what Sofia offers. It finds a way to book or order directly on the site.
Listed by OpenStreetMap · Overture Maps.
Book directly: www.sofiaristoranteroma.it/reservation/ on the business's own site · Menu: www.sofiaristoranteroma.it/menu/
See what an AI agent reads on this site →
40.3 of 65 points available
agent-readiness level, Basic Web Presence
things to fix
Provisional: the action test and the assistant citation check still to come; the hatched part of the bar is not measured yet.
Since the scan of 2026-10-14 (F then): fixed The site could not be read at all. new The server turned AI crawlers away at the door; No structured data for the business; No llms.txt.
Each assistant is asked three questions about this business: the official website, where to book, where to order. The cell says where the answer sent people.
Not asked yet. Assistants are asked about the top-ranked businesses in each city index, and about every claimed business the moment it is claimed; claim this page to be included.
Top 36% of the sites scanned in this index. Just above: Peppone C, Royal art café C. Just below: Al Biondo Tevere C, All'Antico Vinaio C.
| What the top tenth of the index does | Top 10% | This site |
|---|---|---|
| Take bookings or orders on their own domain | 94% | yes |
| Publish the menu or price list as HTML | 94% | yes |
| Let AI crawlers in (no edge block, no robots.txt block) | 77% | no |
| Structured data with address, phone and hours | 35% | no |
| Have an llms.txt | 22% | no |
The top tenth is the 69 best-scoring reachable sites of the index at the last refresh.
| Average score | A or B | This site | |
|---|---|---|---|
| Sites built with Jimdo 600 sites | 38% | 0% | 62% |
| Place | Distance | Grade | Direct booking or ordering |
|---|---|---|---|
| Gregory's Live Jazz & Dinner Club international, italian | 41 m | D | no |
| La Scalinata de Vincenza italian_pizza, italian | 155 m | F | no |
| Crispi 19 italian | 162 m | B | yes |
| Origano Trevi pizza, italian | 171 m | D | no |
| Antica Trattoria Tritone italian | 185 m | F | no |
| alla Rampa italian | 187 m | C | no |
| That's Amore italian | 197 m | D | no |
| La Prosciutteria Roma Trevi - Cantina dei Papi italian | 237 m | C | no |
| DEROMA italian, salad, pasta, grill, pizza | 241 m | B | yes |
| Colline Emiliane italian | 244 m | B | yes |
| Group | Sites | Avg score | A or B | This site |
|---|---|---|---|---|
| Businesses of the same kind in Rome | 688 | 51% | 26% | 62% |
| Sites built with Jimdo in Rome | 2 | 48% | 0% | 62% |
All'oro A 100% · Berberè A 100% · Le Caveau A 100% · Sacco A 100% · l'Archeologia A 100%
With the two fixes above (No structured data for the business; The server turns AI crawlers away at the door) the score would be about 95.4%: 8th of 688 in Rome, up from 250th.
Peers are the businesses of the same index with a reachable site, latest scan of each; the rank uses the index's current ranking.
Add schema.org JSON-LD for the business (Restaurant, LocalBusiness or the closest type, with name, address, telephone and openingHoursSpecification, plus servesCuisine and acceptsReservations). It is the one block every assistant reads first; the fix kit below has it filled in. How to fix it
Live fetches as GPTBot (http_403), ClaudeBot (http_403) were refused while a normal browser got the page. This is usually a hosting or firewall bot list the owner never chose. Ask the host to allow these user agents. How to fix it
Optional today, cheap to add: a short text file describing the business for language models. How to fix it
No one is listed yet to fix this in Rome. List your agency or yourself: free, reviewed by hand.
Builder: jimdo · Front: cloudflare · contact page /chi-siamo-about-us/: nothing new; menu page /menu/: nothing new; booking page /reservation/: nothing new; Cloudflare in front; AI crawlers blocked at the edge
Filled from the business's own record and this scan. Check every line before it goes live; where it goes on jimdo is in the guides.
One block in the page's head, or in the builder's custom-code slot. Replace anything that is wrong. Guide.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Restaurant",
"name": "Sofia",
"url": "https://www.sofiaristoranteroma.it/",
"address": {
"@type": "PostalAddress",
"streetAddress": "Via di Capo le Case 51"
},
"geo": {
"@type": "GeoCoordinates",
"latitude": 41.9039944,
"longitude": 12.4847817
},
"servesCuisine": [
"italian"
],
"acceptsReservations": "https://www.sofiaristoranteroma.it/reservation/",
"hasMenu": "https://www.sofiaristoranteroma.it/menu/",
"potentialAction": [
{
"@type": "ReserveAction",
"target": {
"@type": "EntryPoint",
"urlTemplate": "https://www.sofiaristoranteroma.it/reservation/",
"actionPlatform": [
"http://schema.org/DesktopWebPlatform",
"http://schema.org/MobileWebPlatform"
]
},
"result": {
"@type": "Reservation",
"name": "Reservation"
}
}
]
}
</script>
Save as /llms.txt at the root of the site. Guide.
# Sofia > restaurant (italian) in Via di Capo le Case 51, Rome. ## Book or order directly - Book: https://www.sofiaristoranteroma.it/reservation/ (on the business's own site) - Menu: https://www.sofiaristoranteroma.it/menu/ ## About - Website: https://www.sofiaristoranteroma.it/ - This file follows llms.txt (https://llmstxt.org): a short plain-text description for language models.
Fixed something, or think a check is wrong? Rescans are free. Claiming takes one email to confirm; after that the three assistants are asked about the business right away, the site is rescanned weekly, and you get a message only when the score changes.
Rescan now · Markdown · Dispute a check
Keep watching this business: the assistants asked again every month, with what changed; 2 EUR a month or 19 EUR a year. Deep report: ten questions to each assistant and the places nearby; 5 EUR once. Reading and scanning stay free; these few euros pay for the questions sent to the assistants and for the servers.
The direct record an agent is served for this business, the rails it could transact through, and the agent standards the site publishes.
The links an assistant is sent when it books, orders or reads the menu for this business, where each leads, and when a browser last checked it. The same record is served as JSON at /api/direct/sofiaristoranteroma.it, through the MCP tool direct_record and in the city feed /city/rome/direct.json.
| Path | Link | Status |
|---|---|---|
| Book | www.sofiaristoranteroma.it/reservation/ own site | found by the scan of 2026-10-15, not yet tested |
| Order | none on record | |
| Menu | www.sofiaristoranteroma.it/menu/ html_page | from the scan |
| Hours | none on record | |
| Phone | none on record |
Stable links that count: /go/sofiaristoranteroma.it/book, /go/sofiaristoranteroma.it/menu answer 302 to the links above; the count is the line "followed N direct links" on this page.
Owners: claim the business below to correct this record; the claim email carries the link.
Measured, not scored: the Agent-Direct Score stays at specification v1.0. This says which rails an agent could transact through today, and what the booking or ordering vendor's terms allow.
| UCP profile /.well-known/ucp, Google and Shopify's Universal Commerce Protocol | no |
| ACP discovery document /.well-known/acp.json, OpenAI and Stripe's Agentic Commerce Protocol | no |
| MCP server card /.well-known/mcp/server-card.json | no |
| Ordering vendor that takes agent orders Square, Toast, or a vendor whose terms allow agents | no |
The standards an agent looks for before it reads a page, on the same criteria as Cloudflare's Agent Readiness scan. Not part of the score; each missing item links to a how-to. Cloudflare's report.
| Link headers on the homepage | no how |
| sitemap.xml | yes |
| AI bot rules in robots.txt | yes |
| Content Signals in robots.txt | no how |
| Markdown on Accept: text/markdown | no how |
| llms.txt | no how |
| API catalog (RFC 9727) | no how |
| MCP server card | no how |
| A2A agent card | no how |
| Agent Skills index | no how |
| OAuth or OIDC discovery | no how |
| OAuth protected resource metadata | no how |
| auth.md | no how |
| Web Bot Auth key directory | no how |
| ARD capability manifest | no how |
| DNS-AID records | no how |
| WebMCP tools on the page | no how |
| JSON-LD for the business | no |