agentleFS
Sign inStart free
n24q02mn24q02m/wet/skills/fact-check/SKILL.md

fact-check

Skill19 starsChanged 9 months agoApache-2.0View on GitHub

Verify a claim using adversarial search — find both supporting AND contradicting evidence

We split this skill into the rules it gives an agent

4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 31 edited versions we read keep these.

  • 4

    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:

    Verify a claim by actively searching for BOTH supporting and contradicting evidence.

    line 9
  2. Distinctive:

    All Tier 1-2 sources agree -> Supported (High confidence)

    line 46
  3. Distinctive:

    Only Tier 3-4 sources support, Tier 1-2 contradict -> Unsupported (High)

    line 49
  4. Distinctive:

    Outdated facts: Check publication dates. A 2020 statistic may be wrong in 2026.

    line 58

“Common” is measured: kept by at least 6 in 10 of the 31 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

Fact Check

Verify a claim by actively searching for BOTH supporting and contradicting evidence. Counteracts LLM confirmation bias by enforcing adversarial search.

Steps

  1. Decompose the claim into verifiable sub-claims:
    • Separate factual assertions from opinions
    • Identify the core falsifiable statement
    • Note the domain: scientific, technical, historical, statistical, or current events
  2. Search for SUPPORTING evidence using search:
    • search(action="search", query="[claim as stated]")
    • For scientific claims: search(action="search", query="[claim] research evidence", search_type="academic")
    • Extract 2-3 strongest supporting sources with extract
  3. Search for CONTRADICTING evidence (mandatory — do NOT skip):
    • search(action="search", query="[claim] debunked OR wrong OR myth OR criticism")
    • search(action="search", query="[opposite of claim] evidence")
    • For scientific claims: search(action="search", query="[claim] replication failure OR meta-analysis", search_type="academic")
    • Extract 2-3 strongest contradicting sources with extract
  4. Assess source quality for each piece of evidence:
    • Tier 1: Peer-reviewed journals, official statistics, primary sources
    • Tier 2: Established news outlets, institutional reports, expert analysis
    • Tier 3: Blogs, forums, social media, opinion pieces
    • Tier 4: Anonymous sources, undated content, known biased outlets
    • Discard Tier 4 unless no better sources exist
  5. Produce verdict with structured output:
    • Claim: [exact claim being checked]
    • Verdict: Supported / Partially Supported / Disputed / Unsupported / Insufficient Evidence
    • Confidence: High / Medium / Low (based on source tier and agreement)
    • Supporting evidence: [bullet list with source tier tags]
    • Contradicting evidence: [bullet list with source tier tags]
    • Nuance: [important caveats, context-dependence, or scope limitations]

Decision Tree: Choosing Verdict

  • All Tier 1-2 sources agree -> Supported (High confidence)
  • Tier 1-2 support, some Tier 2-3 contradict -> Partially Supported (Medium)
  • Tier 1-2 sources disagree with each other -> Disputed (note the split)
  • Only Tier 3-4 sources support, Tier 1-2 contradict -> Unsupported (High)
  • Fewer than 2 sources found either way -> Insufficient Evidence (Low)

Common Pitfalls

  • Confirmation bias: Searching only for supporting evidence. The contradicting search in step 3 is MANDATORY.
  • Recency bias: Old but well-cited research may outweigh recent blog posts.
  • Authority bias: A claim from a famous person still needs evidence.
  • Survivorship bias: "X worked for company Y" does not mean X works in general.
  • Outdated facts: Check publication dates. A 2020 statistic may be wrong in 2026.

When to Use

  • Verifying claims before including them in code comments or documentation
  • Checking technical claims ("X is faster than Y", "Z is deprecated")
  • Validating statistics or numerical claims from any source
  • Fact-checking user assumptions before building on them

Download now (ZIP)

  • Other versions38 projects use a version of this.

    Word for word · 1

    1. n24q02m/wet-mcp★ 19100% same

    Edited · 0

    None found.

    Wrote their own version · 37

    1. leonardomso/33-js-concepts★ 67ksame name
    2. THU-MAIC/OpenMAIC★ 41ksame name
    3. garrytan/gbrain★ 31ksame name
    4. elvisun/newsjack★ 1.6ksame name
    5. srid/emanote★ 963same name
    6. akitaonrails/my-skills★ 270same name
    7. Jamie-BitFlight/claude_skills★ 67same name
    8. jezweb/vite-flare-starter★ 49same name
    9. AmethystLuna/embedded-workbench★ 15same name
    10. ArchieSA/skills★ 14same name

    27 more not listed

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • Used alongsideProjects with it also use deep-research, research and 2 more.

    Other files people use alongside this one.

    Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
    • NVIDIASkill

      deep-research

      Queue deep, multi-step research and analysis tasks to the DeepAgents worker. Supports execution depth presets, request-specific rubrics, SubAgent delegation, and RubricMiddleware cross-validation.

      Also in 4 projects that use this

    • NVIDIASkill

      research

      Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.

      Also in 4 projects that use this

    • Microsoft AzureSkill

      humanizer

      Tone enforcement patterns for external-facing community responses

      Also in 3 projects that use this

    • BlockSkill

      frontend-design

      Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.

      Also in 4 projects that use this

  • In this repository7 other files it gives its agents.

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