Claim Grounding

by io.github.truepointmarketingpartners-dotcom

Check answer

check_answer

Read-onlyIdempotent

Fact-check a whole answer: extracts every checkable claim and verifies each against Wikipedia, plus Wikidata structured facts for dates, numbers, quantities, ordinals and records. Returns per-claim verdict, confidence, quote, source URL, sentence span and suggested_correction (quoted only from cited sources when contradicted/outdated), plus an answer-level summary with overall_verdict, counts and problem_sentences. Abstained rather than guessed. Use to audit chatbot or RAG output. Price: $1.00 USDC on Base per call via x402; this MCP call returns payment instructions and POST /v1/check_answer returns the results. Read-only.

Input Schema

Name	Required	Description
`answer`	Yes	The full answer text to fact-check. Each factual sentence is checked separately.
`sources`	No	Optional. Up to 10 sources of your own, used as extra evidence next to the public sources and cited with provider and source_type "caller". The response lists what was used in sources_used.
`sources_only`	No	Optional. true = judge the claim only against your sources (no public-source lookups). Needs a non-empty sources array.

Output Schema

Name	Required	Description
`tool`	Yes	Tool name.
`result`	No	Paid result: { result_id, version, mode, summary: { overall_verdict, counts, problem_sentences }, claims: [{ text, verdict, confidence, citations[], sentence, suggested_correction, next_step? }], recommended_next_steps, next_actions, receipt }.
`status`	Yes	payment_required: pay the HTTP route with x402 to get the result. ok: result is included.
`message`	Yes	Human-readable summary.
`payment`	No	Payment instructions. MCP tools/call does not charge or run the paid work; call the HTTP route with x402.

Schema Changelog

Changes observed during successful MCP inspections.

Changed3 schema fields changedOct 20, 2026

changedOutput schema / properties / result / description

Previous value: -"Paid result: { result_id, version, mode, summary: { overall_verdict, counts, problem_sentences }, claims: [{ text, verdict, confidence, citations[], sentence, suggested_correction }], next_actions, receipt }."New value: +"Paid result: { result_id, version, mode, summary: { overall_verdict, counts, problem_sentences }, claims: [{ text, verdict, confidence, citations[], sentence, suggested_correction, next_step? }], recommended_next_steps, next_actions, receipt }."

addedOutput schema / properties / result / properties / claims / items / properties / next_step

Added value: +{
+  "description": "1.4.8: present only on unsure, low-confidence or unverified high-worthiness claims. A ready-to-send Deep verify request for this claim (optional, paid separately).",
+  "properties": {
+    "body": {
+      "properties": {
+        "claim": {
+          "properties": {
+            "text": {
+              "type": "string"
+            }
+          },
+          "required": [
+            "text"
+          ],
+          "type": "object"
+        }
+      },
+      "required": [
+        "claim"
+      ],
+      "type": "object"
+    },
+    "endpoint": {
+      "format": "uri",
+      "type": "string"
+    },
+    "method": {
+      "const": "POST",
+      "type": "string"
+    },
+    "price_usd": {
+      "type": "number"
+    },
+    "tool": {
+      "const": "deep_verify",
+      "type": "string"
+    },
+    "why": {
+      "type": "string"
+    }
+  },
+  "required": [
+    "tool",
+    "why",
+    "price_usd",
+    "endpoint",
+    "method",
+    "body"
+  ],
+  "type": "object"
+}

addedOutput schema / properties / result / properties / recommended_next_steps

Added value: +{
+  "description": "1.4.8: summary of claim-level next_step suggestions.",
+  "properties": {
+    "count": {
+      "minimum": 0,
+      "type": "integer"
+    },
+    "endpoint": {
+      "format": "uri",
+      "type": "string"
+    },
+    "note": {
+      "type": "string"
+    },
+    "tool": {
+      "type": "string"
+    },
+    "total_price_usd": {
+      "type": "number"
+    }
+  },
+  "required": [
+    "tool",
+    "count",
+    "total_price_usd"
+  ],
+  "type": "object"
+}

Changed2 schema fields changedOct 14, 2026

changedOutput schema / properties / result / description

Previous value: -"Paid result: { result_id, mode, claims: [{ text, verdict, confidence, citations[] }], next_actions, receipt }."New value: +"Paid result: { result_id, version, mode, summary: { overall_verdict, counts, problem_sentences }, claims: [{ text, verdict, confidence, citations[], sentence, suggested_correction }], next_actions, receipt }."

addedOutput schema / properties / result / properties / summary
```
Added value: +{
+  "type": "object"
+}
```

Changed4 schema fields changedOct 13, 2026

addedInput schema / properties / sources

Added value: +{
+  "description": "Optional. Up to 10 sources of your own, used as extra evidence next to the public sources and cited with provider and source_type \"caller\". The response lists what was used in sources_used.",
+  "items": {
+    "additionalProperties": false,
+    "anyOf": [
+      {
+        "required": [
+          "url"
+        ]
+      },
+      {
+        "required": [
+          "text"
+        ]
+      }
+    ],
+    "description": "One source: {url} for a public https page we fetch, or {text, title?, url?} for text you already have.",
+    "properties": {
+      "text": {
+        "description": "Source text to check the claim against (up to 20,000 characters). When given, no fetch is made.",
+        "maxLength": 20000,
+        "minLength": 1,
+        "type": "string"
+      },
+      "title": {
+        "description": "Optional title shown in citations.",
+        "maxLength": 200,
+        "type": "string"
+      },
+      "url": {
+        "description": "Public https URL. Fetched with a 3.5 s timeout and a 512 KB cap; http, file and other schemes, private / loopback / link-local / metadata addresses, IP-literal hosts, custom ports and URLs with credentials are rejected (HTTP 400, not charged). With text, it is only the citation link.",
+        "format": "uri",
+        "maxLength": 2048,
+        "type": "string"
+      }
+    },
+    "type": "object"
+  },
+  "maxItems": 10,
+  "type": "array"
+}

addedInput schema / properties / sources_only

Added value: +{
+  "default": false,
+  "description": "Optional. true = judge the claim only against your sources (no public-source lookups). Needs a non-empty sources array.",
+  "type": "boolean"
+}

addedOutput schema / properties / result / properties / claims / items / properties / citations / description

Added value: +"Evidence quotes with url, title and source_type (wikipedia, wikidata, news, europepmc, fred, sec_edgar, or caller for your own sources)."

addedOutput schema / properties / result / properties / sources_used

Added value: +{
+  "description": "Present when you sent sources: one entry per source with index, kind, url, title, status (used, fetch_failed, empty), chars and reason.",
+  "items": {
+    "type": "object"
+  },
+  "type": "array"
+}

First observedOct 12, 2026

TDQS

A4.2/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already cover the safety profile (readOnly, idempotent, non-destructive, openWorld), so the bar is lower. The description still adds valuable non-obvious behavior: it abstains rather than guessing, and critically that this MCP call returns payment instructions while POST /v1/check_answer returns results, plus the $1.00 USDC/x402 cost. Per-claim return detail is somewhat redundant with the output schema.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose is front-loaded, followed by return contents, usage, and payment mechanics in a logical order. It is dense and slightly run-on with a long enumerated return list, but nearly every clause carries real information.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

An output schema exists, yet the description still adequately signals the response shape; it also documents the source-usage behavior, abstention semantics, and the payment/endpoint split an agent must understand to invoke it correctly. Nothing needed to call it is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% and all three parameters are thoroughly documented in the schema (including fetch timeouts, URL restrictions, char caps). The description adds no parameter-level detail beyond what the schema already provides, so the baseline of 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Fact-check a whole answer') and immediately differentiates scope: it extracts every checkable claim and verifies each against Wikipedia plus Wikidata. The 'whole answer' framing clearly separates it from siblings like verify_claim (single claim) and extract_claims.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives a concrete use case ('Use to audit chatbot or RAG output') and implies the scope condition (whole-answer auditing), which distinguishes it from the single-claim siblings. However, it never names the alternatives (verify_claim, deep_verify, extract_claims) or states when NOT to use it.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Claim Grounding tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.