lenz-fact-check

lenz-fact-check is a skill for Claude Code, Codex from lenzhq/lenz-mcp. It costs 168 tokens per session (1,490 once invoked), scanned A, original, Apache-2.0.

A fact-checking skill that checks factual claims against independent sources on the web through the Lenz service. It breaks text into individual claims and returns an assessment for each one.

In plain words
What is it for?
Use it to verify statements in articles, paragraphs, documents, or datasets, and to select or follow up on claims that need checking.
Why use it?
It helps separate claims that are supported from claims that need correction or more evidence. Lenz must be connected before the checks can run.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to verify statements in articles, paragraphs, documents, or datasets, and to select or follow up on claims that need checking.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/lenzhq/lenz-mcp/lenz-fact-check
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add lenzhq/lenz-mcp --skill lenz-fact-check
Clone the repo
git clone --depth 1 https://github.com/lenzhq/lenz-mcp

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for lenz-fact-check

README.md
[![agentmods](https://agentmods.dev/badge/skills/lenzhq/lenz-mcp/lenz-fact-check/github.svg)](https://agentmods.dev/skills/lenzhq/lenz-mcp/lenz-fact-check)
Your own site
<a href="https://agentmods.dev/skills/lenzhq/lenz-mcp/lenz-fact-check"><img src="https://agentmods.dev/badge/skills/lenzhq/lenz-mcp/lenz-fact-check/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for lenz-fact-check

Your own site · 80×15
<a href="https://agentmods.dev/skills/lenzhq/lenz-mcp/lenz-fact-check"><img src="https://agentmods.dev/badge/skills/lenzhq/lenz-mcp/lenz-fact-check.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 168 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,490 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00168 $0.01490
Opus 5 $0.00084 $0.00745
Sonnet 5 $0.00034 $0.00298
Haiku 4.5 $0.00017 $0.00149

Measured 10d ago against content hash 263b3e7e0cb2, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

lenz-fact-check scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/lenz-fact-check/SKILL.md · 106 lines

How it starts

The opening of the file, as written. The whole thing — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Lenz Fact-Check

Fact-check factual claims against independent web sources using Lenz's hosted MCP tools. Lenz runs a claim through a multi-model pipeline (research → debate → adjudication) and returns a verdict with bucketed confidence. It checks claims against the open web, independent of whatever context the model was given — so it complements groundedness/faithfulness checkers, it does not replace them.

Prerequisite: the Lenz MCP must be connected

This skill drives the Lenz MCP server (https://lenz.io/mcp) and its tools: assess_claim, verify_claim, get_verification, select_claims, ask_followup, check_usage. If those tools are not available, do not try to fact-check by other means — tell the user to connect Lenz first (OAuth for clients that support it, or a free API key), per https://github.com/lenzhq/lenz-mcp, then retry.

Workflow

  1. Extract the atomic claims. Break the input into discrete, individually checkable factual statements — one assertion each. Skip opinions, predictions, recommendations, and subjective statements; Lenz checks facts, not judgments. If there is no checkable factual claim, say so plainly and stop.

  2. Assess each claim with assess_claim (fast, ~10s). It returns a verdict (True / Mostly True / Mixed / Mostly False / False) and a bucketed confidence per claim. If assess_claim reports the claim is ambiguous with candidate readings, pick the reading that matches the user's intent (or ask which they mean), then re-assess that reading.

  3. Escalate to verify_claim only when warranted. verify_claim is a deep, sourced, ~90s investigation that costs an order of magnitude more credits than assess_claim — reserve it for claims that are consequential (health, safety, legal, financial, reputational), came back Mixed or low-confidence from assess_claim, or that the user explicitly wants investigated. Do not spend verify_claim on trivial or clearly-true claims. When you do escalate but the user wants speed or is short on credits, pass depth: "low" — a shallower research pass (fewer sources, faster, the same models) at half the credits. Keep the default standard depth for consequential claims, where breadth of evidence is the point. verify_claim returns a task_id; poll get_verification(task_id) until its status is completed. If it returns needs_input (multiple claims or an ambiguity), use select_claims to choose which claim text(s) to run. To dig further into a finished verify_claim, use ask_followup with its verification_id.

Read the full file on GitHub · 106 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 106 lines · 168 tokens per session scan A 263b3e7e0cb2

Subscribe to this mod's changes

lenz-fact-check is a skill published in the GitHub repository lenzhq/lenz-mcp (0 stars, last pushed 6d ago), licensed Apache-2.0. It adds 168 tokens to every session and 1,490 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.