agentleFS
Sign inStart free
Starlight-IntelligenceStarlight-Intelligence/starlight-agent-skills/skills/research/claim-verification/SKILL.md

claim-verification

Skill1 starsChanged 4 months agoMITView on GitHub

Fact-check a claim against sources and return a verdict with evidence and confidence — supported, refuted, partially-supported, or unverifiable. Use when asked to 'verify', 'fact-check', 'is this true', or before publishing a statistic or scientific claim.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before a claim ships in a post, brief, or video, it should be checked.

    line 14

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Claim Verification

A claim + (optional) sources → a verdict with evidence and a calibrated confidence.

Purpose

Before a claim ships in a post, brief, or video, it should be checked. This skill takes a single claim, gathers or accepts sources, and returns a structured verdict that separates what the evidence supports from what it doesn't — with honest confidence, not false certainty.

When it fires

  • Keyword triggers: verify, fact-check, is this true, confirm, source check, debunk
  • File triggers: claims.md, fact-check/**
  • Command triggers: /claim-verification

Inputs

InputRequiredNotes
claimyesThe exact statement to check.
sourcesnoProvided URLs/quotes; else the skill gathers them.
strictnessnostandard (default) or high (primary sources only).

Workflow

  1. Restate the claim precisely — pin the testable assertion; flag ambiguity. ✅ Check: the claim is falsifiable as stated.
  2. Gather evidence — find/accept sources; prefer primary and authoritative.
  3. Weigh — does the evidence support, refute, or only partly cover the claim? Note source quality and any conflict.
  4. Verdict + confidence — assign one of the four verdicts and a confidence level with a one-line rationale.
  5. Cite — quote the decisive evidence with locations.

Output contract

{
  "claim": "string — restated precisely",
  "verdict": "supported | refuted | partially-supported | unverifiable",
  "confidence": "low | medium | high",
  "evidence": [{ "quote": "string", "source": "string" }],
  "rationale": "string",
  "caveats": ["string"]
}

Tools & MCP

  • Tools: WebSearch, WebFetch, Read.
  • MCP dependencies: optional docs/library MCP for technical claims.

Quality bar

  • Verdict matches the evidence; unverifiable is used honestly.
  • Confidence is calibrated, not defaulted to "high".
  • Every verdict-driving quote is real and located.
  • Source quality is considered, not just source count.

Example

Input: "The James Webb Space Telescope orbits the Earth." → verdict refuted (it orbits the Sun–Earth L2 point), high confidence, with a cited NASA source.

Built on SIP — Starlight Intelligence Protocol Substrate: starlightintelligence.org/protocol v1.1.0 Layers used: [file-contract, attestation, sovereignty] Vertical: starlight-agent-skills · portable capability layer

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free