Claim Grounding

io.github.truepointmarketingpartners-dotcom由 GitHub 帳號 truepointmarketingpartners-dotcom 發布v1.4.7更新於 Oct 25, 2026

Fact-check AI answers with cited sources. Pay per call in USDC on Base via x402.

已驗證Streamable HTTP可網頁執行網頁搜尋與抓取知識與記憶

概覽

AI 產生的概覽

一個遠端 MCP 伺服器,把文字拆成事實性主張並對照公開來源查核,回傳判定、信心分數與引用,按次付費。

可以做什麼
把一段文字或 AI 回答拆成原子事實主張,標註類型、位置與查核價值。對照維基百科查核單一主張,回傳判定、信心分數、理由、引文與來源。逐條查核整段回答,並對日期、數字與紀錄補充維基資料事實。對維基百科、維基資料、Google News、Europe PMC、FRED 與 SEC EDGAR 做多來源深度查核,以非同步工作輪詢結果。
適用情境
當助理產出的回答需要附引用的查核,或使用者想在採信某項主張前確認它被支持、被反駁或已過時時適用。
執行需求
遠端 streamable HTTP 端點;不需帳號或 API 金鑰。每次工具呼叫需透過 x402 以 Base 上的 USDC 付費,由 Coinbase CDP facilitator 結算;個人也可用信用卡結帳。需要能連線該端點及其上游來源的網路。
安裝前請注意
每次工具呼叫都會產生費用(例如 verify_claim 為 1.00 美元,deep_verify 為 0.75 美元),應在用戶端設定最高金額以免超支。僅在回應成功時才結算付款。結果僅限公開英文來源,可能回傳證據不足;不構成法律、醫療或財務建議。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Claim Grounding,將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.truepointmarketingpartners-dotcom%2Fclaim-grounding/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "claim-grounding": {
      "type": "http",
      "url": "https://claim-grounding.prospectintel.workers.dev/mcp"
    }
  }
}

README

TruePoint Prospect Intelligence & Claim Grounding

Pay-per-call APIs for AI agents. Pay USDC on Base per call with x402. No API key, no signup. People can pay by card instead.

  • Prospect Intelligence: work-email lookup and company research for sales agents.
  • Claim Grounding: checks factual claims against public sources and returns a verdict, confidence and citations.

Accuracy results: https://claim-grounding.prospectintel.workers.dev/accuracy (92 of 100 hard claims right on Deep verify, 0 confidently wrong; 38 of 38 found emails exactly right). Read the limitations there too.

30-second quickstart

Option A: Coinbase agentic wallet (awal)

bash
npx awal@latest status          # sign in / check walletnpx awal@latest balance         # needs USDC on Basenpx awal@latest x402 pay https://claim-grounding.prospectintel.workers.dev/v1/verify_claim \  -X POST -d '{"claim":{"text":"The Eiffel Tower is in Paris"}}' --max-amount 1000000

--max-amount is in USDC atomic units (1000000 = $1.00), so you can never pay more than you set.

Option B: any x402 client. Call the endpoint with no payment. You get HTTP 402 with the price in the body and a payment-required header. Sign the payment with your x402 client and retry. Settlement uses the Coinbase CDP facilitator on Base (eip155:8453).

Option C: MCP. Point an MCP client (streamable HTTP) at:

  • https://claim-grounding.prospectintel.workers.dev/mcp
  • https://prospect-intelligence.prospectintel.workers.dev/mcp

tools/list is free. Tool calls return x402 payment instructions; the paid HTTP route does the work.

Option D: pay by card (people). Stripe checkout, no wallet needed:

Card prices are higher than agent prices because Stripe has a $0.50 minimum and per-charge fees.

Prices (USDC per call)

Claim Grounding: https://claim-grounding.prospectintel.workers.dev

MethodPathPriceWhat it does
POST/v1/extract_claims$0.05Split text into checkable claims
POST/v1/verify_claim$1.00Check one claim (Wikipedia)
POST/v1/check_answer$1.00Split an AI answer into claims and check each (Wikipedia)
POST/v1/deep_verify$0.75Check against six sources: Wikipedia, Wikidata, Google News titles, Europe PMC, FRED, SEC EDGAR. Async: poll GET /v1/jobs/:id (free)

Verdicts: supported, contradicted, outdated, mixed, insufficient_evidence.

Prospect Intelligence: https://prospect-intelligence.prospectintel.workers.dev

MethodPathPriceWhat it does
POST/v1/email$0.08, charged only when an address is foundOne work email for a name at a domain, with verification status, source and a send / send_with_caution / verify_first / do_not_send recommendation
POST/v1/basic$0.10research_company
POST/v1/enriched$0.75enrich_prospect
POST/v1/tools$0.75Any one named tool: research_company, research_person, find_decision_makers, analyze_marketing_opportunity, identify_buying_signals, enrich_prospect, score_prospect, generate_prospect_brief
POST/v1/deep$3.00Full prospect brief with a 0–90 score

Sample request and response

Request (real, paid, Oct 12, 2026):

bash
npx awal@latest x402 pay https://claim-grounding.prospectintel.workers.dev/v1/verify_claim \  -X POST -d '{"claim":{"text":"The Eiffel Tower is in Paris"}}'

Response (HTTP 200, shortened with …):

json
{  "mode": "fast",  "job_id": null,  "claim": {    "claim_id": "clm_14b3ffa736fd4fd1",    "text": "The Eiffel Tower is in Paris",    "span": { "start": 0, "end": 29 },    "grounding_required": true,    "verdict": "supported",    "status_detail": "supported",    "confidence": 0.95,    "citations": [      {        "url": "https://en.wikipedia.org/wiki/Eiffel_Tower",        "title": "Eiffel Tower",        "quote": "The Eiffel Tower is a lattice tower on the …",        "retrieved_at": "…",        "provider": "wikipedia"      }    ],    "rationale_short": "…"  },  "why": "…",  "source": "https://en.wikipedia.org/wiki/Eiffel_Tower",  "receipt": { "claim_hashes": ["sha256:…"], "evidence_urls": ["https://en.wikipedia.org/wiki/Eiffel_Tower"], "verdicts": ["supported"], "priced_usd": 1, "created_at": "…" }}

Paid on Base: tx 0x839f…a901.

The unpaid first call (real, live) returns HTTP 402:

json
{  "x402Version": 2,  "error": "Payment required",  "accepts": [{    "scheme": "exact",    "network": "eip155:8453",    "amount": "80000",    "asset": "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",    "payTo": "0x485FfC037f02dDDc154A454a9cA89C4B1b29e25b",    "maxTimeoutSeconds": 300,    "extra": { "name": "USD Coin", "version": "2" }  }]}

(80000 = $0.08 for /v1/email. The asset is USDC on Base.)

Errors

StatusWhenCharged?
200SuccessYes
402No or invalid payment. Body and payment-required header list what to payNo
400Bad input, e.g. missing claim.text, or mode: "deep" sent to verify_claim (use /v1/deep_verify). On email: missing domain or namesNo
404Email lookup found no address (no_result), unknown path, or unknown job idNo
405Non-POST request to /mcpNo
409Card payment already used for a lookup (one payment = one lookup)No
422Email provider rejected the input (e.g. domain has no mail records)No
451Email lookup: the person opted out of data processing. Do not contact or storeNo
502 / 503Upstream source or email provider unavailable. Safe to retry laterNo

x402 only settles payment on a 2xx response, so failed calls cost nothing.

What these tools don't do

  • Not legal, medical, financial or compliance advice. Not a certification.
  • English only. Public sources only; no paywalled or private documents.
  • Can return insufficient_evidence. Very recent news may lag.
  • No guarantee an email is delivered or read. No personal/private emails, no phone numbers.
  • Decision-makers are unknown unless a public source names them. Nothing is invented.

Discovery

/docs · /.well-known/x402 · /openapi.json · /llms.txt · /.well-known/mcp/server-card.json on each host. Listed on agentic.market.

Worker source stays private; this repo is the public surface for agents and builders.

Support

來源:README.md,提交 2ab74d0

工具

4

extract_claims

Extract claims

Split a text (for example an AI-generated answer) into atomic factual claims. Each claim includes span, grounding_required, claim_type (number|date|quantity|quote|named_entity_fact|comparison|opinion|prediction|other), check_worthiness (high|medium|low), and the source sentence. Hedges are stripped, pronouns are resolved across sentences, and compound sentences are split. Questions, greetings, opinions and predictions are marked grounding_required false. Use before verify_claim. Price: $0.05 USDC on Base per call via x402; this MCP call returns payment instructions and POST /v1/extract_claims returns the claims. Read-only.

輸入參數text必填

verify_claim

Verify claim

Check one factual claim against public Wikipedia sources and return a verdict (supported, contradicted, outdated, mixed or insufficient_evidence), a confidence score, a short reason, the supporting quote and the source URL. Says insufficient_evidence instead of guessing. Price: $1.00 USDC on Base per call via x402; this MCP call returns payment instructions and the paid HTTP route POST /v1/verify_claim returns the verdict. For a multi-source check use deep_verify. Read-only.

輸入參數claim必填sourcessources_only

check_answer

Check answer

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.

輸入參數answer必填sourcessources_only

deep_verify

Deep verify

Use when a claim is high-stakes, or when check_answer or extract_claims marks it unsure (their next_step gives the exact request). Verifies up to three claims against Wikipedia, Wikidata, Google News, Europe PMC, FRED and SEC EDGAR; returns verdict, confidence, evidence for and against with URLs, and flags contradictions. Async: poll GET /v1/jobs/{job_id} free. $0.75 USDC on Base via x402. Read-only.

輸入參數answer必填sourcessources_only

版本歷史

2
  1. v1.4.7最新Oct 15, 2026
  2. v1.4.4Oct 12, 2026

評分

已滿足適用檢查項的 90%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分

    請在儲存庫或註冊中心條目中宣告授權條款。