Every change Wellknown observed on this MCP server, newest first, with what it was before and what it became. Tool-surface changes carry the definition diff. Nothing here is edited after the fact.
Changed the definition of "rankings"
⟨40 unchanged words⟩ with fewer than 100 scanned sites are listedunranked.unranked, and those with fewer than four in five of their own websites scanned are listed as partly scanned. Page: https://directindex.org/rankings?kind=.
Added "menu_prices" (19 tools before, 20 now)
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 water, read from the menus on restaurants', cafes' and bars' own websites, with the number of menus behind each figure, the day they were read and how many prices a hand check found right. Covers 13 European cities so far (Paris, Madrid, Berlin, Milan, Lisbon, Vienna and others); without a city it lists them. Figures are menu prices, not what a bill comes to.
Added "city_facts" (18 tools before, 19 now)
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 menu, take bookings or orders directly or grade A or B; the share that declare an English version, show prices, link to WhatsApp, show an old copyright year or load a consent manager; the middle visible price and the page weight. Each with its base, rank and medians. Page: https://directindex.org/city/{slug}#sites and https://directindex.org/stats.
Changed the definition of "assistant_destinations", "assistant_recommendations", "city_changes" and 15 more
⟨11 unchanged words⟩ city together","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"limit":{"default":25,"description":"How ⟨3 unchanged words⟩ 200,"minimum":1,"type":"integer"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"
Changed the definition of "order_deep_report"
Only numbers and dates differ: this server embeds live figures in the definition, so the text moves without the contract changing.
Order the paid deep report (195 EUR, one-off) for a business in a ⟨97 unchanged words⟩
Certificate recorded, valid to 2027-01-11
Authorization not required
Changed the definition of "assistant_destinations", "assistant_recommendations", "city_changes" and 15 more
before
—after
Assistant destinationsbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Assistant destinations"}before
—Added "assistant_destinations", "assistant_recommendations", "city_changes" and 8 more; changed the definition of "find_direct_places" (7 tools before, 18 now)
⟨19 unchanged words⟩ with the direct booking or ordering link foreach.each (the owner's corrected link where the business is claimed), the vendor behind it, verified_at when a browser reached the step before confirming, the vendor's terms for agents when noted, the hours, and follow links (https://directindex.org/go/{domain}/book, /order, /menu) that count. Use it to recommend places a person can ⟨18 unchanged words⟩ for example pizza, sushi, vegan, francesinha. Optional filters: lat, lon and radius_m for places near a point; open_now for places open at this moment in the city's zone (unknown hours stay in, marked null); verified_only for links a browser verified; vendor for one booking or ordering vendor (slug as on https://directindex.org/vendors). The same lists exist as pages at https://directindex.org/city/{slug}/direct, /direct/book, /direct/order and /direct/{want}.; the full record of one place: direct_record.
Added "get_deep_report" and "order_deep_report"; changed the definition of "find_direct_places" (5 tools before, 7 now)
⟨54 unchanged words⟩ , for example pizza, sushi, vegan, francesinha. The same lists exist as pages at https://directindex.org/city/{slug}/direct, /direct/book, /direct/order and /direct/{want}.
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).
Order the paid deep report (19 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 places, the fix plan with the kit to paste, what an agent reads on the site, a page and a PDF. Returns the order token and a Stripe checkout link: give the link to the person to pay (nothing is charged until they do); the report is written in about fifteen minutes after payment and also emailed. Poll get_deep_report with the token.
Unknown → Live
First tool surface recorded: 5 tools (server version 0.2.0)
https://directindex.org/mcp (mcp_streamable_http) — from mcp_registry, with the record
Showing the latest 13 events. The API returns up to 500 and filters by kind: ?kind=tool_surface_changed
⟨8 unchanged words⟩ see list_cities","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"intent":{"description":"A question key to ⟨9 unchanged words⟩ empty for all","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["city","context","llm_model"],"type":"object"}
⟨8 unchanged words⟩ see list_cities","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"days":{"default":30,"maximum":365,"minimum":1,"type":"integer"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["city","context","llm_model"],"type":"object"}
⟨6 unchanged words⟩ example lisbon","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"limit":{"default":50,"maximum":500,"minimum":1,"type":"integer"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["city","context","llm_model"],"type":"object"}
⟨21 unchanged words⟩ example porto","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"kind":{"default":"restaurants","description": ⟨14 unchanged words⟩ clinics","tours"],"type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["a","b","context","llm_model"],"type":"object"}
⟨12 unchanged words⟩ list of countries","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"domain":{"description":"Registered domain, for example belcanto.pt","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["domain","context","llm_model"],"type":"object"}
⟨8 unchanged words⟩ see list_cities","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"lat":{"description":"Latitude of the person; ⟨13 unchanged words⟩ "minimum":1,"type":"integer"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"lon":{"description":"Longitude of the person ⟨118 unchanged words⟩ type":"string"}},"required":["city","context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"order":{"description":"The 32-character order token from order_deep_report","type":"string"}},"required":["order","context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"domain":{"description":"Registered domain, for example belcanto.pt","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["domain","context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"domain":{"description":"The business's registered domain ⟨41 unchanged words⟩ da or cs","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["domain","email","context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"kind":{"default":"restaurants","description": ⟨17 unchanged words⟩ tours","all"],"type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"url":{"description":"Website address, for example yourbusiness.com","type":"string"}},"required":["url","context","llm_model"],"type":"object"}
⟨21 unchanged words⟩ or berlin-cafes-bars","type":"string"},"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"finding":{"description":"Finding id, for ⟨27 unchanged words⟩ ,"tours"],"type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"page":{"default":1,"minimum":1, ⟨22 unchanged words⟩ zenchef, toast","type":"string"}},"required":["context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"slug":{"description":"Vendor slug, for example zenchef","type":"string"}},"required":["slug","context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"},"vendor":{"description":"Vendor slug as it ⟨6 unchanged words⟩ opentable, square","type":"string"}},"required":["context","llm_model"],"type":"object"}
{"properties":{"context":{"description":"Explain in 15-25 words, in third person, why this tool is called and how it supports the user's goal. For analytics only. You MUST describe only the abstract purpose of the tool call. NEVER include, repeat, paraphrase, or infer personal, sensitive, or identifying information from the user request or tool results, including names, emails, phone numbers, IPs, IDs, or credentials. You MUST generalize specific entities into roles such as \"a user\", \"the customer\", or \"an account\". Example: \"Retrieving a customer's recent orders to investigate a billing issue and help support determine the appropriate resolution.\"","type":"string"},"conversation_id":{"description":"Pass the exact conversation_id from the server's previous response, unchanged. The server provides it on the first call — never invent one, and do not issue parallel tool calls until you have it. Keep passing the same conversation_id for the rest of the conversation, including after later user messages or on a different task; do not reset it when the user starts a new request.","type":"string"},"domain":{"description":"Registered domain, for example belcanto.pt","type":"string"},"llm_model":{"description":"The exact model identifier you (the assistant) are running as, taken from your system prompt or environment (e.g. \"claude-opus-4-8\", \"gpt-5.2\"). Used for analytics only. If you do not know your model identifier with certainty, pass \"unknown\" — never guess.","type":"string"}},"required":["domain","context","llm_model"],"type":"object"}
after
Assistant recommendationsbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Assistant recommendations"}before
—after
City changesbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"City changes"}before
—after
City leaderboardbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"City leaderboard"}before
—after
Compare citiesbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Compare cities"}before
—after
Country summarybefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Country summary"}before
—after
Direct recordbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Direct record"}before
—after
Find direct places⟨165 unchanged words⟩ {want}; the full record of one place:direct_record.is available by domain.
before
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Find direct places"}before
—after
Get deep reportbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Get deep report"}before
—after
Get site reportbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Get site report"}before
—after
List citiesbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"List cities"}before
—after
Order deep report⟨66 unchanged words⟩ the order token and a Stripe checkout link:give the link tofor the person to pay (nothing is charged ⟨8 unchanged words⟩ in about fifteen minutes after payment and alsoemailed.emailed,Pollandget_deep_reportitswithstatus and text can be read by the token.
before
—after
{"destructiveHint":false,"idempotentHint":false,"openWorldHint":true,"readOnlyHint":false,"title":"Order deep report"}before
—after
Rankingsbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Rankings"}before
—after
Scan websitebefore
—after
{"destructiveHint":false,"idempotentHint":false,"openWorldHint":true,"readOnlyHint":false,"title":"Scan website"}before
—after
Search placesbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Search places"}before
—after
Vendor reportbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Vendor report"}before
—after
Vendor termsbefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"Vendor terms"}before
—after
What assistants saybefore
—after
{"destructiveHint":false,"idempotentHint":true,"openWorldHint":false,"readOnlyHint":true,"title":"What assistants say"}⟨8 unchanged words⟩ see list_cities","type":"string"},"lat":{"description":"Latitude of the person; with lon, only places within radius_m","type":"number"},"limit":{"default":10,"maximum":25,"minimum":1,"type":"integer"},"lon":{"description":"Longitude of the person","type":"number"},"open_now":{"default":false,"description":"Only places open right now by their OpenStreetMap hours, in the city's zone; places with unreadable hours stay in with open_now null","type":"boolean"},"query":{"description":"What the person wants ⟨7 unchanged words⟩ words. Empty for any.","type":"string"},"radius_m":{"default":1000,"description":"Metres around lat, lon","maximum":50000,"minimum":50,"type":"integer"},"vendor":{"description":"Only places on this booking or ordering vendor, by slug: zenchef, sevenrooms, toast ...","type":"string"},"verified_only":{"default":false,"description":"Only links a browser verified: an agent reached the step before confirming","type":"boolean"},"want":{"default":"any", ⟨22 unchanged words⟩
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 and overall, and the headline shares (direct, marketplace, the top marketplace). Pages: https://directindex.org/city/{slug}/assistants and https://directindex.org/assistants for every city.
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 place's own site, and the direct booking or ordering link where the place has one. Pages: https://directindex.org/city/{slug}/recommended and /recommended/{question}.
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 business. Each change links the business's report. Also a page (https://directindex.org/city/{slug}/changes), an Atom feed (/changes.xml) and JSON (/api/city/{slug}/changes?days=30).
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 assistants' direct share where probed, and the month-by-month numbers. Takes index slugs or city names (lisbon, Porto, berlin-cafes-bars). Page: https://directindex.org/compare/{a}/{b}.
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 country with an index. Pages: https://directindex.org/country/{code} and /countries.
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 (OpenStreetMap form and schema.org, with open_now), phone, address, the assistants' latest verdicts, how often agents read and followed it this month, the score, and the source and date of every fact. Corrected by the owner where the business is claimed. Also GET https://directindex.org/api/direct/{domain}.
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 fewer than 100 scanned sites are listed unranked. Page: https://directindex.org/rankings?kind=.
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 Agent-Direct Scores first, each with its report link. Free, no contacts. Page: https://directindex.org/search.
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, example sites, and the vendor's terms for agents when a note exists. Vendor slugs as on https://directindex.org/vendors (zenchef, sevenrooms, thefork, toast ...).
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. Use it before sending a person to book or order through a vendor on an agent's behalf.
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 for which the business was named, and one sentence on why an assistant sent people elsewhere. For businesses in a city index that have been asked about; see the report at https://directindex.org/r/{domain}.