Lenz Fact Check

Name: Lenz Fact Check
Author: lenzhq

@lenzhq

リポジトリ

Lenz Fact Check について

概要はまだありません

基本情報

カテゴリ

その他

トランスポート

stdio

公開者

lenzhq

投稿者

Pavel

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "lenz": {
      "type": "http",
      "url": "https://lenz.io/mcp"
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Lenz Fact Check?

Lenz Fact Check is a remote MCP server that takes a factual claim and runs it through a multi-model pipeline (framing, research, debate, panel review, conclusion) against independent sources on the open web, returning a verdict with a directional confidence level. It is designed for any MCP client and complements retrieval or groundedness checkers.

How to use Lenz Fact Check?

No installation is required. Point your MCP client to https://lenz.io/mcp with transport type http. Authentication is via OAuth or a free Lenz API key (Authorization: Bearer lenz_…). Use tools like assess for a fast verdict (~5–10s) or verify for a deep investigation (~90s) that returns a task_id to poll.

Key features of Lenz Fact Check

Multi-model pipeline (framing → research → debate → panel review → conclusion)
Fast verdict via 3-model panel (assess) in ~5–10 seconds
Deep investigation (verify) for high-stakes claims in ~90 seconds
Follow-up grounded questions (ask) on completed verifications
Resolves multi-claim or clarification interrupts (select)
Remote, hosted service – no local install needed

Use cases of Lenz Fact Check

Verifying factual claims against independent web sources without relying on the model’s training data
Performing quick sanity checks on statements during research or content generation
Conducting thorough, multi-step investigations for high-stakes or sensitive claims
Asking clarifying follow-up questions after a completed verification
Monitoring API usage and remaining quota via the check_usage tool

FAQ from Lenz Fact Check

How do I connect my MCP client to Lenz Fact Check?

Point your client to https://lenz.io/mcp with transport type http and use OAuth or a free Lenz API key (Authorization: Bearer lenz_…).

What kind of verdicts does it return?

Verdicts are directional, not absolute – confidence is bucketed and includes a caveat. The assess tool returns a quick verdict; verify returns a more detailed investigation.