A source-vs-spin analysis tool that cross-references media claims with primary evidence.

Install

mkdir -p .claude/skills/refute && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/12330" && unzip -o skill.zip -d .claude/skills/refute && rm skill.zip

Installs to .claude/skills/refute

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Counter-narrative analysis tool. Takes a mainstream media article about the DPRK/Korea and systematically deconstructs it — tracing the source chain, identifying opinion laundered as fact, and presenting what primary sources actually say. Outputs a publishable "Source vs. Spin" rebuttal.
288 charsno explicit “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Advanced

Key capabilities

  • →Extract factual claims from media articles
  • →Identify attribution and evidence types for each claim
  • →Trace the source chain of how a story propagated
  • →Find primary source information from DPRK/Korean sources
  • →Document missing context or omissions in articles
  • →Generate a structured 'Source vs. Spin' rebuttal

How it works

The skill deconstructs a media article by extracting claims, identifying evidence types, tracing source chains, finding primary source reality, and documenting missing context, then generates a structured rebuttal.

Inputs & outputs

You give it
A mainstream media article URL or text about the DPRK/Korea
You get back
A publishable 'Source vs. Spin' rebuttal comparing reported narratives against primary sources

When to use refute

  • →Verify media claims about DPRK
  • →Identify circular reporting in news
  • →Expose unverified anonymous source claims

About this skill

/refute — Source vs. Spin Counter-Narrative Engine

Takes a mainstream media article and produces a structured, evidence-based rebuttal by comparing what was reported against what primary sources actually say. When the reporting holds up, it says so and writes no rebuttal.

When to Use

  • An MSM article makes claims about the DPRK that misrepresent or lack primary sources
  • You want to expose circular reporting (outlets quoting each other as "confirmation")
  • You need a side-by-side comparison of narrative vs. reality
  • You want to document the opinion-laundering pipeline for a specific story

Not for NPSI / npsi.ca work. NPSI reads Korean primary sources through npsi-korean-translation and takes no advocacy position.

How to Invoke

Paste an article URL or text and say /refute or "refute this".

Analysis Pipeline

Phase 1: Read It Straight (deep-reading-analyst)

Before anything else, run the deep-reading-analyst skill on the article at Deep level. It steelmans the author, grades every claim, maps the source chain, and runs competing hypotheses and inversion (what would make each claim false?). Keep its grades; this skill does not re-grade:

  • A Verified Fact · B Supported Inference · C Expert Opinion
  • D Anonymous Source (and claims stated as fact with no source at all, marked UNSOURCED)
  • E Editorial Framing · F Circular Citation

The gate. Count the load-bearing claims, the ones the headline and the lede rest on. Go on to Phase 2 only when more than half of them are graded D, E or F. Otherwise the reporting holds up: the verdict is Sound. Hand over deep-reading-analyst's analysis, add any missing context as a note, and stop. Do not write or publish a rebuttal.

Phase 2: Trace the Source Chain

Map how this story propagated:

ORIGIN: [Who first made this claim? Think tank? Single analyst? Anonymous official?]
   ↓
FIRST PICKUP: [Which wire service or outlet first reported it?]
   ↓
AMPLIFICATION: [Who quoted the first pickup? How many outlets?]
   ↓
ECHO CHAMBER: [How many outlets now cite each other as "multiple sources confirm"?]

UNIQUE INDEPENDENT SOURCES: [actual count — usually 1]

Phase 3: Find the Primary Record

Search the primary sources on this topic and keep two separate lists.

What each government said (its own words, labelled as its claim):

  • KCNA statements
  • Rodong Sinmun articles
  • Official statements from every government involved
  • Historical context from the Korean perspective, labelled as that perspective

What the independent record shows:

  • Actual text of treaties, agreements, or laws referenced
  • Documents, imagery, UN Panel of Experts reports, court records, data

A KCNA or Blue House statement is primary evidence of what that government said, not of what happened. Grade it on the same scale as the article.

If no primary source exists on this specific topic, state that clearly — don't fabricate one. Instead, note what context is missing.

Phase 4: Identify What's Missing

Document what the article omits (state the omission; do not guess why):

  • Historical context (what provoked the action/statement?)
  • The other side's stated position (in their own words)
  • Relevant actions by every party that preceded the event
  • Scale/proportion (is this being inflated?)
  • Track record (how many similar predictions were wrong?)

Phase 5: Generate the Rebuttal

Output Format

# Source vs. Spin: [Topic]

**Article**: [Title] — [Publication], [Date]
**Author**: [Name]

---

## The Narrative

> [Key quote from the article that captures the framing]

## Claim-by-Claim Breakdown

### Claim 1: "[exact claim]"
- **Source**: [who said this]
- **Evidence grade**: [A-F, from Phase 1; D marked UNSOURCED where no source is given]
- **What the government said**: [KCNA/official, labelled as that government's claim]
- **What the independent record shows**: [documents, imagery, UN Panel of Experts, court records, data — or "no independent record"]
- **What's missing**: [omitted context]

### Claim 2: "[exact claim]"
[same structure]

### Claim 3: "[exact claim]"
[same structure]

---

## Source Chain

| Step | Source | What They Said | Based On |
|------|--------|---------------|----------|
| Origin | [name/org] | [claim] | [their evidence] |
| Wire | [outlet] | [reframing] | [cited origin] |
| Amplifier 1 | [outlet] | [further reframing] | [cited wire] |
| Amplifier 2 | [outlet] | [now "confirmed"] | [cited amplifier 1] |

**Unique independent sources**: [count]
**Total outlets reporting as "confirmed"**: [count]

---

## What the Primary Sources Say

[Each government's statements with dates and full context, labelled as its claims; then the independent record, or "no independent record"]

---

## Missing Context

1. [Critical context omitted]
2. [Historical precedent ignored]
3. [Proportionality not addressed]

---

## Verdict

[2-3 sentence assessment: How reliable is this article? What is the actual evidence level? What should a reader take away?]

---

*Analysis produced for juche.org — Primary sources over opinion. Evidence over narrative.*

Quality Standards

  1. Never fabricate primary sources. If we don't have a DPRK source on this topic, say so. The strength of this tool is honesty — we don't need to make things up.

  2. Steelman before attacking. Present the MSM article's strongest interpretation before dismantling it. This builds credibility.

  3. Be precise with language. Don't say "they lied" — say "this claim is graded D (UNSOURCED) because no source is cited." Let the evidence speak.

  4. Distinguish levels of bad reporting. Not every MSM article is wrong, and not every wrong one is propaganda. Some hold up (sound), some are lazy (missing context), some are biased (selective framing), some rest on sources that turn out not to exist (fabricated). Grade accordingly.

  5. Always provide the primary source in full context. Don't cherry-pick KCNA quotes either. If we criticize them for it, we can't do it ourselves.

  6. Date everything. Claims without dates are claims without context.

  7. Critique the article, never impute motive to the named journalist. The grades describe the reporting, not the person who wrote it.

Credibility Tiers for MSM Articles

TierDescriptionResponse
SoundThe reporting holds up: grades A-C carry the load-bearing claimsSay so and do not publish a rebuttal
LazyMissing context, incomplete reporting, no primary sources consultedProvide the missing context
BiasedSelective framing, loaded language, but factual core is presentReframe with full context
CircularMultiple outlets citing each other as confirmation of a single sourceExpose the source chain
UnverifiedLoad-bearing claims rest on anonymous sources only, or on no source at allDocument what is and is not sourced
FabricatedA cited source or quote is shown not to existDocument the evidence that it does not exist
PropagandaDeliberately deceptive framing, with evidence of intent (shown, not inferred from the framing); without that evidence the tier is BiasedFull deconstruction with evidence

Sound is decided at the Phase 1 gate. Every other tier applies only after the gate, when D-F grades dominate.

When not to use it

  • →When the article is not about the DPRK/Korea
  • →When the user wants to fabricate primary sources
  • →When the user wants to attack the article without steelmanning

Limitations

  • →Never fabricates primary sources
  • →Requires a mainstream media article about the DPRK/Korea
  • →Focuses on counter-narrative analysis

How it compares

This skill systematically analyzes media narratives against primary sources to expose spin, which is more rigorous and evidence-based than a subjective critique.

Compared to similar skills

refute side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
refute (this skill)08moNo flagsAdvanced
linkedin-sales-navigator-alt235moNo flagsIntermediate
market-research-reports3811moReviewAdvanced
annas-archive-ebooks2210moReviewBeginner

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

linkedin-sales-navigator-alt

OneWave-AI

Build targeted prospect lists by analyzing LinkedIn profiles, extracting job titles, companies, locations, and recent activity. Identifies decision-makers, tracks job changes for warm outreach, and enriches contact data. Use when users need to find prospects, build lead lists, or track decision-maker movements.

23226

market-research-reports

davila7

Generate comprehensive market research reports (50+ pages) in the style of top consulting firms (McKinsey, BCG, Gartner). Features professional LaTeX formatting, extensive visual generation with scientific-schematics and generate-image, deep integration with research-lookup for data gathering, and multi-framework strategic analysis including Porter's Five Forces, PESTLE, SWOT, TAM/SAM/SOM, and BCG Matrix.

38162

annas-archive-ebooks

ratacat

Use when needing to look up book content, find a book by title/author, download an ebook, or reference material from a published book. Triggers on book lookups, ebook downloads, "find the book", "get the PDF/EPUB of". Downloads produce PDF/EPUB/MOBI files - use ebook-extractor skill to convert to text.

22177

content-trend-researcher

alirezarezvani

Advanced content and topic research skill that analyzes trends across Google Analytics, Google Trends, Substack, Medium, Reddit, LinkedIn, X, blogs, podcasts, and YouTube to generate data-driven article outlines based on user intent analysis

30126

trend-research

gtmagents

Use when identifying cultural moments, audio trends, memes, or creator

8113

competitive-intelligence

anthropics

Research your competitors and build an interactive battlecard. Outputs an HTML artifact with clickable competitor cards and a comparison matrix. Trigger with "competitive intel", "research competitors", "how do we compare to [competitor]", "battlecard for [competitor]", or "what's new with [competitor]".

1553

Search skills

Search the agent skills registry