investigate-journalism-skills

A structured framework for AI agents to revise a prior verdict, classification, hypothesis, or analysis when new evidence emerges — calibrated against both under-correction (anchoring on the prior) and over-correction (anchoring on novelty or social pressure). Use whenever the task involves updating a previously reached conclusion in light of new information, reconsidering a verdict after primary sources are fetched, deciding whether a contested claim should change classification, integrating a late-breaking piece of evidence into an existing investigation or peer review, or any request to "update this", "revise this", "rethink this in light of X", "does this new evidence change anything", "should I change my mind". Pair with whichever analytical skill produced the prior — this skill takes that output and produces a calibrated revision in its format.

A structured framework for AI agents to analyse text for logical fallacies, cognitive biases, rhetorical manipulation, and other forms of untruthful reasoning.

A compact framework for analyzing problems, claims, proposals, or beliefs by decomposing them to their irreducible truths and rebuilding reasoning from the ground up. Use whenever the user wants to dissect a claim, audit an argument, challenge conventional wisdom, redesign a solution from scratch, debug a decision, or asks any variant of "is this actually true?", "why is it done this way?", "what are we assuming?", "break this down from first principles", "is the reasoning sound?", or "what would this look like if we started from zero?". Trigger even when the user doesn't say "first principles" — phrases like "rethink this", "challenge my thinking", "stress-test this idea", or "is this conventional wisdom correct?" should activate it.

Micro-skill for recording gut feelings, hunches, anomaly signals, and "something feels off" reactions as pre-evidence hypotheses before article review, investigations, OSINT, claim audits, rhetoric audits, peer review, scientific fact classification, or belief revision. Use only as a Phase 0 hunch register; intuition may guide tests and searches but never supports a verdict.

A structured framework for AI agents to critically analyze events, detect deception, and develop well-reasoned alternative hypotheses.

A standalone orchestrator for reviewing journalistic articles, news reports, investigations, op-eds presented as reporting, newsletters, and media explainers for accuracy, sourcing, framing, omissions, headline-body alignment, quote context, evidentiary support, and public-interest reliability. Use when the user asks to review, audit, fact-check, critique, verify, assess bias in, or evaluate whether a journalistic article is fair, accurate, misleading, well-sourced, or publication-ready. Route underlying scientific papers to peer-review and individual empirical claims to scientific-fact-classification when those layers become load-bearing.

A structured framework for AI agents to plan, execute, and verify open-source intelligence (OSINT) investigations using only publicly available information (PAI). Use whenever the user wants to investigate or build a profile on a person, organisation, account, domain, IP, image, video, location, vehicle, vessel, aircraft, or event using public web data — including phrases like "find out about", "investigate", "who is behind", "verify this image/video", "where/when was this taken", "what's the digital footprint of", "trace this username/email/domain", "is this account real", "build a profile on", or any request to systematically gather and corroborate online information about a target. Pair with the investigative-reasoning skill when the goal extends to hypothesis construction or narrative challenge — this skill handles gathering and verification mechanics; investigative-reasoning handles analytical reasoning on top.

A standalone orchestrator for rigorous, citation-verified peer review of scientific papers, manuscripts, preprints, studies, systematic reviews, methods papers, and research reports. Use when the task is to evaluate whether a paper's methods, statistics, causal claims, citations, reproducibility, and conclusions support its stated contribution; verify that cited sources say what the paper claims; assess pre-registration, transparency, and conflicts of interest; or produce a referee-style report with Fatal / Major / Minor findings and an explicit recommendation. Route article-level reporting about a paper to journalistic-article-review, individual claim status to scientific-fact-classification, source/network checks to osint-research or investigative-reasoning, rhetoric to fallacy-bias-and-manipulation-analysis, and later corrections to belief-revision.

A structured framework for AI agents to weigh, recognise, and classify claims along the spectrum from objective fact to assumption, opinion, conjecture, and unfalsifiable belief — and to specify the conditions under which a claim qualifies as scientifically warranted. Use whenever the task involves separating facts from assumptions, judging the evidential status of a claim, deciding whether something is 'objective', auditing reasoning for unsupported leaps, or assigning a calibrated confidence level to a statement.

Turn a finished, warrant-labelled analysis into a publishable Dutch markdown article. Use when the user asks to write/publish a completed analysis as an article ("schrijf een artikel", "maak er een publicabel stuk van", "publiceer dit", "turn this into a piece"). This skill writes only; it does not research. It preserves the warrant chain: no new load-bearing facts, no stronger phrasing than the analysis supports, weak claims hedged and attributed.