fabioc-aloha/alex_plug_in · Archived
journalism
News writing, investigative reporting, source verification, editorial standards, and fact-checking for journalists.
Review carefully
Installs
1
Stars
5
Security
No audit
Freshness
Archived
Updated
—
First seen Jun 12, 2026
Installation
Project
Global
Entire repo
Entire repo (global)
$
npx skills add fabioc-aloha/alex_plug_in --skill journalism
Copy
SKILL.md
Journalism Skill
News writing, investigative reporting, source verification, editorial standards, and fact-checking for journalists.
Core Principle
Journalism's job is to tell the truth as completely and fairly as possible. Every story serves the public's right to know. Accuracy is non-negotiable — speed is secondary.
News Writing Fundamentals
The Inverted Pyramid
Lead (Lede) — Most important facts: Who, What, When, Where, Why, How
Key details — Context, quotes, supporting evidence
Background — History, related events, lesser details
The reader should get the essential story from the first paragraph alone.
Lead Types
Type
Use When
Example Pattern
Hard news
Breaking/spot news
WHO did WHAT at WHERE on WHEN
Delayed
Feature/narrative
Scene-setting, character introduction
Summary
Complex stories
Synthesize multiple elements
Anecdotal
Human interest
Start with one person's experience
Question
Rarely (avoid)
Only when genuinely provocative
AP Style Quick Reference
Rule
Correct
Incorrect
Numbers < 10
five people
5 people
Numbers ≥ 10
15 people
fifteen people
State abbreviations
Calif., Texas
CA, TX (except datelines)
Titles before name
President Biden
president Biden
Titles after name
Joe Biden, president
Joe Biden, President
Time
3 p.m.
3 PM, 3:00 p.m.
Dates
March 9, 2026
March 9th, 2026
Source Verification
Source Hierarchy
Tier
Source Type
Reliability
Verification Need
1
Primary documents
Highest
Authenticate provenance
2
Direct witnesses/participants
High
Corroborate with second source
3
Expert analysis
Medium-High
Confirm credentials, check for conflicts
4
Secondary reports (other media)
Medium
Verify independently before citing
5
Anonymous sources
Variable
Editor approval, corroboration required
6
Social media posts
Low
Verify identity, context, authenticity
Two-Source Rule
Every factual claim should be confirmed by at least two independent sources
Independent = not derived from the same original source
Exceptions: official documents, on-record statements, direct observation
Fact-Check Tracker
story: "City council approves budget"
deadline: "2026-04-14 18:00"
claims:
- claim: "Budget is $4.2 billion"
sources:
- type: "Primary document"
ref: "Budget proposal PDF"
verified: true
- type: "Official statement"
ref: "Finance director quote"
verified: true
status: "confirmed"
- claim: "Largest increase in 10 years"
sources:
- type: "City records"
ref: "Historical budget data"
verified: true
- type: "Secondary"
ref: "2016 Observer article"
verified: false
status: "pending verification"
Source Attribution
Attribution
Use When
On the record
Default — named source, direct quotes
On background
Information usable, source described but not named
Deep background
Information usable, no attribution at all
Off the record
Cannot be used in any form (establish BEFORE conversation)
Always : Clarify attribution terms before the conversation, not after.
Fact-Checking Protocol
Pre-Publication Checklist
☐ All names spelled correctly (verify against official sources)
☐ All numbers verified (dates, statistics, amounts, addresses)
☐ All quotes accurate (audio/transcript confirmed)
☐ Claims attributed to sources or verified independently
☐ Headline accurately reflects story content
☐ Context provided for statistics (base rate, time period, methodology)
☐ Opposing viewpoints sought and included where relevant
☐ Potential defamation reviewed (truth, public interest, fair comment)
Statistical Literacy
Red Flag
Question to Ask
Percentage without base
Percent of what? (10% of 10 vs. 10% of 10M)
Correlation claimed as causation
What confounders exist?
Cherry-picked time period
What does the full dataset show?
Survey results without methodology
Sample size? Margin of error? Who funded it?
"Studies show" without citation
Which study? Peer reviewed? Replicable?
Investigative Reporting
Investigation Framework
Hypothesis — What do you suspect, and why?
Document trail — FOIA requests, court records, corporate filings, financial disclosures
Human sources — Insiders, whistleblowers, affected parties, experts
Data analysis — Public databases, leaked datasets, data journalism tools
Pattern recognition — Connect dots across documents, timelines, relationships
Confrontation — Give the subject fair opportunity to respond
Legal review — Pre-publication review for defamation, privacy, source protection
Public Records Toolkit
Record Type
Source
Corporate filings
SEC EDGAR, state SOS databases
Property records
County assessor/recorder
Court records
PACER (federal), state court portals
Campaign finance
FEC, state election commissions
Government contracts
USAspending.gov, SAM.gov
Nonprofit finances
IRS Form 990 (ProPublica Nonprofit Explorer)
Environmental data
EPA Enforcement, Toxic Release Inventory
Ethical Standards
SPJ Code of Ethics (Four Principles)
Seek truth and report it — Accuracy, context, transparency about methods
Minimize harm — Balance public interest against potential damage
Act independently — Avoid conflicts of interest, refuse gifts
Be accountable — Acknowledge mistakes, correct promptly
Correction Protocol
Severity
Action
Minor factual error
Correction appended, original text updated
Significant factual error
Editor's note at top, original text updated
Fundamental error affecting thesis
Retraction or major revision with prominent notice
All corrections
Date-stamped, transparent about what changed
Digital & Multimedia Journalism
Verification of Digital Content
Reverse image search — Google Images, TinEye to check origin
EXIF data — Check photo metadata for date, location, device
Geolocation — Cross-reference landmarks, shadows, signs with claimed location
Video verification — Frame analysis, audio consistency, compression artifacts
Social media verification — Account age, posting history, network analysis
Data Journalism Principles
Show your work: publish methodology and datasets when possible
Visualizations must not misrepresent data (avoid truncated axes, misleading scales)
Acknowledge limitations of the data
Peer review analysis before publication
AI in Journalism — Guardrails
AI can assist with: document analysis, data processing, transcription, translation
AI must never: fabricate quotes, generate fake sources, write opinion as news
All AI-assisted reporting must be disclosed to editors and readers
AI-generated images must never be used as documentary evidence
Every AI-generated draft must be verified line by line against source material
The journalist remains solely responsible for accuracy