Skip to content
Back to skills

Fact Checking

ASecurity

Verify factual claims (theorem statements and hypotheses, definitions and conventions, physical constants, data, names, dates, history) against the learner's own textbooks and official/reference sources with source_lookup, then cite them in the notes or mark them unverified. Use before writing lessons, solutions, reinforcement notes, or when the learner asks "is this true / where does it come from".

  • 3 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 18, 2026
developmentpythonrustgospringapidocumentation

Works with

  • api

Security analysis

A100/100

Scanned November 18, 2026

npx -y skills add Lelpoco2/learning-curve --skill fact-checking --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Fact Checking?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Fact Checking
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/lelpoco2-fact-checking/badge)](https://www.skillsdirectory.com/skills/lelpoco2-fact-checking)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: fact-checking
description: Verify factual claims (theorem statements and hypotheses, definitions and conventions, physical constants, data, names, dates, history) against the learner's own textbooks and official/reference sources with source_lookup, then cite them in the notes or mark them unverified. Use before writing lessons, solutions, reinforcement notes, or when the learner asks "is this true / where does it come from".
---

# Fact checking

Calculations are verified with `math_check` (skill `math-verification`). Everything else
that could be wrong goes through `source_lookup`. Goal: every statement in a note is either
derived in the note, verified by computation, or backed by a cited source — or clearly
marked as unverified.

## What to check

| Check always | Usually no need |
|---|---|
| Exact statement and **hypotheses** of a theorem (continuity vs differentiability, open vs closed interval, finite vs infinite dimension…) | Steps you derive in the note itself |
| Definitions and **conventions** that vary between books (sign conventions, $\mathbb N$ with or without 0, normalisations, units) | Pure algebra/calculus results (use `math_check`) |
| Physical constants, measured data, numeric facts | Your own analogies and intuitive explanations (they are not claims) |
| Names, dates, who proved what, history anecdotes | |
| Formulas you do not derive (special functions, tables, identities) | |
| Anything you are not sure about, or the learner questions | |
| Programming: syntax, API signatures, default values, error types, behaviour that changed between versions — check the official documentation for the learner's version (`url`), and run it (`code_run`) | |

## Source order (first that answers wins)

1. **Learner's sources** (`source: local`): their textbooks, lecture notes, syllabus in
   the vault Sources folder. Their course's convention is the one to follow.
2. **Official / standard references**:
   - Physical constants → `codata` (NIST CODATA, official values with uncertainty).
   - Special functions, identities, integrals → `dlmf` (NIST DLMF).
   - Integer sequences → `oeis`.
3. **Reference encyclopedias**: `eom` (Encyclopedia of Mathematics, EMS/Springer),
   `wikipedia` (prefer `lang: en`, usually more complete; use the notes language for
   terminology), `nlab` (advanced/abstract math, category theory).
4. **Literature**: `arxiv` (papers, original results), `crossref` (exact bibliographic
   data, DOIs). `url` for a specific official page (university course page, standards body,
   the official documentation of a programming language or library: docs.python.org,
   cppreference.com, docs.oracle.com, developer.mozilla.org, doc.rust-lang.org…).

`source: auto` = learner's sources + Wikipedia. Use specific sources for constants,
special functions and sequences.

## How to check

1. Formulate a precise query (key terms, not the whole sentence).
2. Read the excerpt. The claim is verified only if the excerpt states it **with the same
   hypotheses and conventions**. If it only states something similar, it is not verified.
3. Two independent sources for important or surprising claims (e.g. a theorem's exact
   hypotheses, a historical attribution).
4. **Conflicts**: if sources disagree (often conventions), follow the learner's source,
   mention the alternative in a `[!note]` callout ("Some books define … instead"), and
   never silently mix conventions.
5. Not found / sources unreachable: do not invent a citation. Keep the statement only if
   you are confident, and mark it unverified (below), or rephrase it more carefully.

## Write it in the note

- Footnote right after the statement: `… is continuous on $[a,b]$.[^1]`
- `## Sources` section at the end of the note (heading translated into the notes language)
  with the footnotes, using the citation from `source_lookup`:
  - `[^1]: [[Sources/Analisi 1/Bramanti.pdf#page=212|Bramanti, p. 212]]`
  - `[^2]: "Mean value theorem", Wikipedia (en), https://en.wikipedia.org/wiki/Mean_value_theorem (accessed 2026-11-09)`
- Add the source to the note's `sources:` property (short form: file link or URL).
- Unverified statement:
  ```
  > [!warning] Not verified
  > The attribution of this result to … could not be checked against a reliable source.
  ```
- Never cite a source you did not actually open with `source_lookup` in this workspace.

## In chat

When the learner asks "is this true?" or "where does it come from?", answer with the
evidence: short quote/paraphrase, source, link. If the learner's book says something
different from what you said, the book wins: correct the note (and say so).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in development →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.