claims
What AI says about the business.
Claims AI made about the business in one check of a paid order (default: the latest scheduled check). list (scan:read): each grouped claim with its polarity, how many answers made it, which AIs, one short quote, whether the person marked it wrong and when the fix went live. Writes (scan:write), with claim and polarity exactly as list gives them: mark_wrong (undo true takes the mark off) puts it in About your business under "What AI gets wrong about you"; fixed_on (YYYY-MM-DD within 30 days, empty clears) is the day the correction went live. facts (scan:read): the facts check of that check, included in every paid order: each value AI or a listing stated that doesn't match the facts the person gave (phone, address, hours, prices, services, areas, name), with one short quote, the AI, the date, the phrase, the person's own value and the day they saved it, sites_cited_in_that_answer, first and last seen. mark_fact (scan:write; flag_id from facts, mark wrong, right or clear): wrong confirms it; right means the person's own fact is out of date. Only the person decides: never mark one they didn't.
Who can use it
- Access
- A paid order.
- Scope
scan:readfor reading; each write action's scope is below.- Changes anything
- Its write actions do; see below.
Actions
Pass one in action. Without it: list.
| action | Scope | What it does |
|---|---|---|
list | scan:read | What AI said about the business in one check (default: the latest scheduled check). |
facts | scan:read | The facts check of one check (default: the latest scheduled check), included in every paid order: each value AI or a listing stated that doesn't match the facts the person gave on their fact sheet. Compared by code; nothing is called wrong until the person marks it. |
mark_wrong | scan:write | A write action: the description above says what it changes. |
fixed_on | scan:write | A write action: the description above says what it changes. |
mark_fact | scan:write | A write action: the description above says what it changes. |
Inputs
| Name | Type | Required | Values and limits |
|---|---|---|---|
action | string | No | list, facts, mark_wrong, fixed_on, mark_fact. Default list |
order_id | string | Yes | An order id from get_dashboard |
check_id | string | No | |
claim | string | No | 0 to 300 characters |
polarity | string | No | positive, negative, neutral |
undo | boolean | No | For mark_wrong: take the mark off |
fixed_on | string | No | For fixed_on: YYYY-MM-DD, or empty |
flag_id | string | No | For mark_fact: a flag_id from action facts |
mark | string | No | wrong, right, clear. For mark_fact |
What it answers
Every answer has status and, where useful, message in plain words.
| Field | Meaning |
|---|---|
check | The check read: check_id and date. |
counts | negative, positive, neutral, and in how many answers. |
claims | Each grouped claim: claim, polarity, answers, ais, one short quote (at most 200 characters, never the whole answer), values, marked_wrong and fixed_on. |
names_only | Answers that only named the business, with no claim. |
enabled, ran (facts) | Whether the facts check is on, and whether this check was compared with the person's facts. |
facts_used (facts) | The person's own facts the check used: key, label, value and saved_on (the day they saved it). |
flags (facts) | Each value AI stated that doesn't match: flag_id, topic (phone, address, hours, price, service, area), value, seen_in_this_check, first_seen, last_seen, mark (wrong, right or null), yours (key, value, saved_on), line, and answers: answer_id, check_id, ai, date, phrase, one short quote (at most 400 characters) and sites_cited_in_that_answer (site, url, in_sentence, your_site, competitor). |
listing_flags (facts) | The same for the Google listing (name, address, phone, hours) and the Trustpilot and Tripadvisor names: platform, read_on and link. |
matched, notes, not_checked (facts) | Answers per topic that matched; plain notes (a listing with no hours, a name close enough not to flag); why something wasn't checked (no fact given, hours we couldn't read, the run's cap). |
Errors
| status | What it means |
|---|---|
invalid_input | order_id isn't an order id. Take one from get_dashboard. |
not_found | No order with that id on this account. |
invalid_input | An action that isn't in the list, or mark_fact without a flag_id from facts and a mark of wrong, right or clear. |
not_found | mark_fact: no such flag on this order. |
forbidden | The credential lacks the tool's scope, or an agent's owner has no ScanFree account yet (the owner signs in on the website once). |
Input validation error | An argument doesn't match the schema below. The call fails with MCP error -32602, naming the field. |
Errors before the tool runs (a missing key, too many requests) are on the MCP page.
Example
Example shape; values are illustrative.
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "claims",
"arguments": {
"action": "list",
"order_id": "00000000-0000-4000-8000-000000000002"
}
}
}{
"status": "ok",
"order_id": "00000000-0000-4000-8000-000000000002",
"check": {
"check_id": "00000000-0000-4000-8000-000000000003",
"date": "2026-11-18"
},
"counts": {
"negative": 1,
"positive": 3,
"neutral": 2,
"in_answers": 6
},
"names_only": [],
"claims": [
{
"claim": "Open 24 hours",
"polarity": "positive",
"answers": 2,
"ais": [
"Perplexity"
],
"quote": "…",
"marked_wrong": false,
"fixed_on": null
}
],
"note": "…"
}