fact-checker
Skill6 starsChanged 8 months agoMITView on GitHub
Use this skill when verifying factual claims, checking accuracy of statements, or assessing the credibility of information. Trigger phrases: 'fact-check this', 'is this true', 'verify this claim', 'check if this is accurate'. Do NOT use for subjective opinions, legal or medical advice, or claims requiring real-time data or breaking news.
21 other projects use a version of this.
Adopted by
Most recently taken up by
- aditya-vermax/skillstheir own version, first seen Nov 8, 2026
- BrickyStudios/ai-fact-checker-claude-skilltheir own version, first seen Nov 8, 2026
- weininghui/skillstheir own version, first seen Oct 30, 2026
- daymade/claude-code-skillstheir own version, first seen Oct 19, 2026
- yasdelayu/agent-x-cctheir own version, first seen Oct 13, 2026
- IbrahimSobh/AISkillstheir own version, first seen Oct 9, 2026
Used with
Projects with it also use qa-expert, prompt-master and 6 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.
belokonmSkillqa-expert
Quality assurance specialist focusing on test strategy, quality processes, and comprehensive testing methodologies
Also in 3 projects that use this
ttmouseSkillprompt-master
提示词主控 - 智能选择合适的领域skill并生成提示词,支持自动领域分类和调度
Also in 3 projects that use this
NVIDIASkilldeep-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
Microsoft AzureSkillhumanizer
Tone enforcement patterns for external-facing community responses
Also in 4 projects that use this
VercelSkillfind-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Also in 3 projects that use this
GoogleSkillgrill-me
A relentless interview to sharpen a plan or design.
Also in 3 projects that use this
Other versions21 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 21
- daymade/claude-code-skills★ 1.5ksame name
- IbrahimSobh/AISkills★ 9same name
- Bartosh16/claude-skills★ 3same name
- BrickyStudios/ai-fact-checker-claude-skill★ 3same name
- SkillMedev/academic-researcher★ 3same name
- ethicalcapital/skills★ 2same name
- seaneoliver/claude-code-starter★ 2same name
- suchwow-sysadmin/claude-code-factchecker★ 2same name
- aditya-vermax/skills★ 1same name
- arielnishri-svg/claude-skills★ 1same name
11 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.
In this repository68 other files it gives its agents.
- dataset-curatorSkill
- eval-designerSkill
- model-comparatorSkill
- prompt-engineerSkill
- business-analystSkill
- competitor-analystSkill
- market-researcherSkill
- meeting-facilitatorSkill
- okr-plannerSkill
- pitch-deck-writerSkill
- product-managerSkill
- ux-researcherSkill
- api-designerSkill
- architecture-designerSkill
- code-reviewerSkill
- debuggerSkill
- documentation-writerSkill
- refactorerSkill
- regex-masterSkill
- security-auditorSkill
- sql-expertSkill
- test-writerSkill
- chart-builderSkill
- dashboard-designerSkill
- data-analystSkill
- data-cleanerSkill
- excel-expertSkill
- sql-analystSkill
- color-paletteSkill
- design-critiquerSkill
- frontend-designSkill
- ux-writerSkill
- ci-cd-helperSkill
- docker-expertSkill
- kubernetes-helperSkill
- monitoring-setupSkill
- flashcard-makerSkill
- lesson-plannerSkill
- quiz-creatorSkill
- tutorSkill
- docxSkill
- pdfSkill
- pptxSkill
- xlsxSkill
- budget-plannerSkill
- financial-analystSkill
- invoice-generatorSkill
- contract-reviewerSkill
- legal-summarizerSkill
- terms-of-serviceSkill
- decision-makerSkill
- habit-builderSkill
- journalingSkill
- note-takerSkill
- task-plannerSkill
- citation-formatterSkill
- literature-reviewerSkill
- summarizerSkill
- web-researcherSkill
- academic-essaySkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 9 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 19 edited versions we read keep these.
9
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
- Distinctive: line 64
Do other independent sources corroborate or contradict this?
- Distinctive: line 68
Assign a verdict with confidence level: Use a clear rating:
- Distinctive: line 171
Always restate the claim in precise, neutral terms before evaluating—many claims are vague or bundled
- Distinctive: line 174
Use the most direct, primary source available—don't cite a news article if you can cite the study it covers
- Distinctive: line 176
When a claim is partially true, explain exactly which part is true and which is not
- Distinctive: line 190
When checking statistics, always ask: who collected this data, when, from what population, using what methodology?
- Distinctive: line 192
Use reverse image search for photos and videos being presented as evidence of recent events
- Distinctive: line 194
If a claim seems too perfectly suited to confirm someone's beliefs, apply extra scrutiny—motivated reasoning produces motivated …
- Distinctive: line 195
Document your sources as you go.
“Common” is measured: kept by at least 6 in 10 of the 19 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.
Files
Fact Checker
Overview
This skill provides a structured approach to evaluating factual claims—assessing whether a statement is accurate, partially accurate, misleading, or false, and explaining why with evidence and reasoning. Good fact-checking goes beyond a binary true/false verdict: it identifies the precise claim being made, locates the best available evidence, accounts for context and nuance, and rates confidence appropriately. The output is a clear, evidence-backed assessment that helps readers understand not just whether something is accurate, but why it matters.
When to Use
- Verifying statistics, quotes, or factual assertions in articles or documents
- Checking whether a viral social media claim holds up to scrutiny
- Assessing whether historical facts cited in writing are accurate
- Auditing factual claims in a draft before publication
- Evaluating whether scientific findings are being accurately reported
- Confirming whether attributed quotes are real and in context
- Vetting claims in speeches, presentations, or marketing copy
When NOT to Use
- Evaluating subjective opinions or value judgments ("X policy is better than Y")
- Providing legal advice or legal interpretations of statutes
- Providing medical diagnoses or treatment recommendations
- Verifying real-time data such as live stock prices, current weather, or breaking news
- Assessing internal business claims that require proprietary data access
- Deciding which of two contested scientific theories is definitively correct (use
literature-reviewerskill for contested scientific debates)
Quick Reference
| Task | Approach |
| Identify the claim | Isolate the precise factual assertion—strip out opinion and framing |
| Primary sources | Seek original studies, official records, or direct quotes over secondary reports |
| Verdict labels | True / Mostly True / Mixed / Mostly False / False / Unverifiable |
| Context matters | Accurate statistics can still mislead if context is stripped away |
| Quotes | Check original source; confirm attribution, date, and surrounding context |
| Statistics | Verify source, date, sample size, and whether the stat is being applied correctly |
| Confidence level | State confidence (High / Medium / Low) based on source quality and evidence volume |
Instructions
- Isolate the precise claim. Restate the claim in neutral, specific terms before evaluating it. Many claims contain embedded assumptions or vague language that must be unpacked. "Studies show coffee causes cancer" is not a single claim—it bundles together a specific substance, a causal mechanism, and a category of disease. Break compound claims into individual checkable assertions.
- Assess the claim type. Determine whether this is:
- A factual assertion (a specific, verifiable statement about the world)
- A statistical claim (requires checking source, date, methodology, and applicability)
- A quote attribution (requires locating the original source and verifying wording and context)
- A causal claim (requires evaluating whether evidence supports causation or only correlation)
- A historical claim (requires consulting authoritative historical records)
- Identify the best sources. Rank sources by reliability:
- Tier 1: Peer-reviewed research, official government data, primary documents
- Tier 2: Established news organizations with editorial standards, expert consensus statements
- Tier 3: Reputable secondary sources, encyclopedias, expert interviews
- Avoid: Anonymous posts, sites with known bias and no editorial accountability, single-source claims without corroboration
- Evaluate the evidence. Apply these checks:
- Does the original source actually say what the claim implies?
- Is the data current and applicable to the specific context?
- Are there important caveats or limitations the claim omits?
- Do other independent sources corroborate or contradict this?
- Assess context and framing. A claim can be technically true but deeply misleading. Check: Is the statistic cherry-picked from a broader dataset? Is the quote taken out of context? Does the claim imply causation from a correlation? Does it generalize from a narrow study to a broad population?
- Assign a verdict with confidence level. Use a clear rating:
- True: Accurate and properly contextualized
- Mostly True: Accurate in substance but missing important nuance or context
- Mixed: Contains both accurate and inaccurate elements
- Mostly False: Misleading framing or significant inaccuracies, though a kernel may be accurate
- False: Contradicted by the best available evidence
- Unverifiable: Cannot be confirmed or denied with available sources Also note your confidence level (High / Medium / Low) based on the quality and quantity of evidence.
- Write the fact-check clearly. Structure the output as: (1) the claim as stated, (2) the verdict up front, (3) the evidence and reasoning, (4) any important caveats or context. Attribute every piece of evidence to a named source. Avoid weasel words—be direct about what the evidence does and does not show.
Examples
Example 1: Statistical Claim Fact-Check
Input: "Fact-check this claim: 'The Great Wall of China is the only man-made structure visible from space.'"
Output:
Claim: "The Great Wall of China is the only man-made structure visible from space."
Verdict: FALSE Confidence: High
Evidence and Reasoning:
This is one of the most persistent myths in popular culture, but it is contradicted by direct observation, physics, and astronaut testimony.
What astronauts actually report: Multiple astronauts have specifically addressed this claim. Chinese astronaut Yang Liwei, the first Chinese person in space, stated after his 2003 mission that he could not see the Great Wall. NASA astronaut Ed Lu confirmed this in 2004. Far from supporting the myth, the people best positioned to verify it have consistently denied it.
The physics: The Great Wall is, on average, 5–8 meters wide—narrower than a human hair as seen from the International Space Station (roughly 400 km altitude). Visibility at that scale would require impossibly acute vision. NASA's Earth Observatory notes that the wall "would be as difficult to see as a human hair from two miles away."
What IS visible: Many man-made features are visible from low Earth orbit under good conditions, including highways, airports, cities, reservoirs, and agricultural fields—all of which are wider than the Wall. From the Moon (385,000 km away), no man-made structure of any kind is visible.
Origin of the myth: The claim appears as early as 1932 in Ripley's Believe It Or Not and was repeated in reference books for decades before being widely debunked.
Sources: NASA Earth Observatory; Chinese National Space Administration statements; Liu Yang and Yang Liwei mission reports (2003, 2012).
Example 2: Quoted Statistic Fact-Check
Input: "Someone cited this statistic in a presentation: 'We only use 10% of our brains.' Is this true?"
Output:
Claim: "Humans only use 10% of their brains."
Verdict: FALSE Confidence: High
Evidence and Reasoning:
This claim has no basis in neuroscience and is contradicted by multiple independent lines of evidence.
Brain imaging data: Functional MRI and PET scan studies show that virtually all brain regions are active over the course of a day. While not every neuron fires simultaneously (which would cause a seizure), there is no dormant 90% waiting to be unlocked. A 2014 University of Western Ontario study explicitly tested this myth and found activity throughout the brain during normal function.
Evolutionary evidence: The human brain accounts for approximately 2% of body weight but consumes 20% of the body's energy at rest. Evolution would not sustain such a metabolically expensive organ if 90% were unused.
Clinical neurology: Brain damage from strokes, tumors, or injury to virtually any region produces measurable deficits. If 90% were unused, damage to most areas should have no effect—which is not observed.
The myth's origin: No credible scientific source has been identified as the origin. It has been falsely attributed to Albert Einstein (no record of him saying it exists), and may derive from misunderstandings of early neuroscience research distinguishing glial cells from neurons, or from self-help literature of the early 20th century.
The accurate picture: Different regions are specialized for different functions, and humans use all of them—though not all simultaneously. Expertise and learning strengthen specific neural pathways, but this is optimization, not "unlocking" unused capacity.
Sources: Barry Beyerstein, "Whence Cometh the Myth that We Only Use 10% of our Brains?" (1999); Neuroimaging research, University of Western Ontario (2014); Scientific American (multiple articles).
Example 3: Nuanced/Mixed Verdict
Input: "Fact-check: 'Violent crime in the US has been increasing every year for the past decade.'"
Output:
Claim: "Violent crime in the US has been increasing every year for the past decade."
Verdict: MOSTLY FALSE Confidence: Medium
Evidence and Reasoning:
The claim is contradicted by the overall trend but contains a kernel of truth related to recent years.
The long-term trend: FBI Uniform Crime Report data and Bureau of Justice Statistics victimization surveys both show that violent crime in the US peaked around 1991 and declined dramatically through the 2010s. In 2019, violent crime rates were near 30-year lows. The claim that crime has been increasing every year for the past decade is flatly wrong as a description of the 2013–2023 period.
Where it gets complicated: There were notable increases in homicide specifically in 2020 and 2021—likely related to COVID-19 pandemic disruptions—that drew significant media attention. Some cities also saw increases in certain crime categories. However, these were not "every year" and did not represent a consistent decade-long trend. Preliminary 2022–2023 data suggest homicides declined again in many cities.
Measurement caveats: Crime statistics are imperfect. The FBI's transition to a new reporting system (NIBRS) in 2021 created a gap in data from non-reporting agencies that complicates year-over-year comparisons. Reported crime and experienced crime (per victimization surveys) sometimes diverge based on reporting rates.
Verdict summary: The claim exaggerates recent increases and ignores the 30-year decline. Saying violent crime "has been increasing every year for the past decade" is not supported by the data. More accurate: violent crime fell steadily through the mid-2010s, spiked in 2020–2021 (particularly homicide), and has since partially reversed.
Sources: FBI Uniform Crime Report 2022; Bureau of Justice Statistics National Crime Victimization Survey 2022; Brennan Center for Justice crime data analysis.
Best Practices
- Always restate the claim in precise, neutral terms before evaluating—many claims are vague or bundled
- Lead with the verdict, then provide the reasoning—don't make readers wade through evidence to find your conclusion
- Distinguish correlation from causation explicitly when the claim implies causation
- Use the most direct, primary source available—don't cite a news article if you can cite the study it covers
- Quantify your confidence—saying "High confidence" vs "Low confidence" signals important uncertainty to readers
- When a claim is partially true, explain exactly which part is true and which is not
- Be direct: avoid hedging so much that the verdict is unclear
Common Mistakes
- Fact-checking the wrong claim: Evaluating a related but different claim than the one actually made
- Stopping at "technically true": A claim can be technically accurate but still deeply misleading—evaluate framing too
- Single-source verification: Confirming a claim from only one source, especially when that source is the same one cited in the original claim
- False balance: Treating fringe views as equivalent to scientific consensus in the name of "both sides"
- Scope mismatch: A study from a specific population or country doesn't verify a universal claim
- Ignoring the date: An accurate statistic from 10 years ago may be outdated and no longer accurate
- Verdict without reasoning: Stating a verdict without showing the evidence and logic behind it
Tips & Tricks
- For quote attributions, search the exact phrase in quotation marks and look for the primary source, not secondary reports
- When checking statistics, always ask: who collected this data, when, from what population, using what methodology?
- The original study abstract (and especially limitations section) often reveals important caveats that secondary reports omit
- Use reverse image search for photos and videos being presented as evidence of recent events
- High-confidence verdicts require at least two independent, high-quality sources
- If a claim seems too perfectly suited to confirm someone's beliefs, apply extra scrutiny—motivated reasoning produces motivated evidence-seeking
- Document your sources as you go; reconstructing a citation trail after the fact is tedious and error-prone
Related Skills
- web-researcher
- literature-reviewer
- summarizer
- proofreader
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
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.

