A structured framework for AI agents to critically analyze events, detect deception, and develop well-reasoned alternative hypotheses. Use this skill whenever an agent needs to investigate an event skeptically, assess official narratives, identify logical red flags, apply Cui Bono / follow-the-money reasoning, detect logical fallacies, or build a stepwise hypothesis. Trigger this skill for tasks like: analyze who benefits from X, find red flags in this narrative, develop an alternative theory for Y, apply critical thinking to this event, check this explanation for logical fallacies, or any investigative / deception-detection task.
A rigorous methodology for AI agents to investigate events, challenge narratives, and construct well-reasoned alternative hypotheses. Derived from classical critical thinking, detective reasoning, and the scientific method.
Only use this skill when explicitly requested by the user. Do not apply this framework proactively or assume investigative intent. Wait for a clear instruction such as:
Before investing effort, determine whether a conspiracy theory is warranted:
If none of these apply, no conspiracy theory is needed. State this clearly.
Scan the official narrative for these categories of red flags:
Does the alleged perpetrator have a coherent reason to act?
Could the alleged perpetrator actually execute the operation?
Was the alleged perpetrator in a position to act?
Does the official account violate known laws of physics, logistics, or common sense?
Flag count matters: One anomaly = coincidence. Two = suspicious. Three or more = investigate seriously. (Adapted from Ian Fleming's principle.)
Before building theories, systematically gather raw material. The quality of a theory is only as good as the evidence beneath it.
Start broad, then narrow. Use layered searches:
Establish the baseline — What is the official narrative?
"[event name]" official statement OR press release OR reportFind the cracks — Where does the official story face criticism?
"[event name]" inconsistencies OR anomalies OR contradictions OR questions"[event name]" investigation OR inquiry OR whistleblowerFollow the money — Who profited?
"[event name]" financial OR contract OR stock OR beneficiary OR profit[key actors] net worth OR funding OR donations OR lobbyingFind alternative analyses — What do independent researchers say?
"[event name]" alternative explanation OR independent analysis OR criticism"[event name]" site:substack.com OR site:independentresearcher.comHistorical parallels — Has something like this happened before?
[pattern type] historical examples OR precedent OR similar eventEvaluate every source before using it. Assign a trust tier:
| Tier | Source Type | Trust Level | Notes |
|---|---|---|---|
| 1 | Primary documents (court filings, leaked docs, official reports, FOIA releases) | High | Still check for authenticity |
| 2 | Academic papers, peer-reviewed research | High | Check funding sources |
| 3 | Established investigative journalism (e.g. ICIJ, ProPublica, Der Spiegel) | Medium-High | Still has editorial bias |
| 4 | Mainstream news outlets | Medium | Cross-reference; known to amplify official narratives |
| 5 | Independent journalists, substack, blogs | Medium-Low | Depends heavily on sourcing quality |
| 6 | Anonymous sources, social media posts, forums | Low | Never use alone; require corroboration |
| 7 | AI-generated content or unsourced claims | Very Low | Require independent verification |
Key rule: A claim is only as strong as its weakest supporting source. Always trace claims back to their primary source.
Mainstream and alternative media can both be captured or biased. Deliberately seek sources from opposing viewpoints:
When using web search tools during investigation:
DO:
✓ Use specific named entities: people, places, organizations, dates
✓ Search for primary documents: "[event] report filetype:pdf"
✓ Search for financial connections: "[actor A] [actor B] funding OR investment"
✓ Use date ranges to find pre-narrative coverage: "[event] before:[date]"
✓ Search for deleted/archived content: site:web.archive.org "[url]"
✓ Cross-check facts across at least 3 independent sources
✓ Look for the original source when a claim is cited
DON'T:
✗ Accept top search results as truth — they may be SEO-optimized official narratives
✗ Rely on a single source for a key claim
✗ Ignore sources that contradict the emerging theory
✗ Confuse high publication frequency with credibility
✗ Treat Wikipedia as a primary source (it is a secondary source, often edited)
For each event under investigation, build a fact file:
FACT FILE: [Event Name]
────────────────────────────────────────────
Date & Location:
Key actors named in official narrative:
Official explanation summary:
Primary sources reviewed: [list with URLs]
Contradicting sources reviewed: [list with URLs]
Key claims that lack primary sourcing:
Financial flows identified:
Key figures' backgrounds & connections:
Timeline anomalies:
Physical/forensic evidence available:
Evidence that is missing or was destroyed:
Witnesses: names, statements, credibility notes
Watch for signs that information may be managed or planted:
Ask: Who benefits from the outcome?
For each potential actor, assess:
| Actor | Benefit Type | Benefit Description | Plausibility |
|---|---|---|---|
| Party A | Financial / Power / Strategic | … | High/Med/Low |
| Party B | … | … | … |
Benefit types to consider:
Important: Cui Bono gives a starting point, never direct proof. Multiple actors can benefit simultaneously ("killing multiple birds with one stone").
Build a suspect matrix:
| Suspect | Motive ✓/✗ | Means ✓/✗ | Opportunity ✓/✗ | Score |
|---|---|---|---|---|
| Actor A | ✓ | ✓ | ✓ | 3/3 |
| Actor B | ✓ | ✗ | ✓ | 2/3 |
| Actor C | ✗ | ✓ | ✗ | 1/3 |
Rank suspects by score. Focus theory development on the highest-scoring actors.
Use the appropriate reasoning mode for the available evidence:
General rule → specific conclusion
"Controlled demolitions always produce free-fall collapses. Building X collapsed in free-fall. Therefore Building X may have been a controlled demolition."
⚠️ A deductively valid argument can still be wrong if the general premise is false.
Specific observations → general pattern
"Past color revolutions were organized by intelligence services. This event resembles a color revolution. Therefore intelligence service involvement is plausible."
⚠️ Inductive conclusions can always be overturned by exceptions.
Choose the most plausible explanation from competing hypotheses.
Preferred when empirical evidence is scarce. Select the theory that explains the most facts with the fewest unexplained residuals.
"If it looks like a duck, swims like a duck, and quacks like a duck — it probably is a duck." Apply when multiple surface characteristics match a known pattern.
Before finalizing a theory, screen for these biases:
| Bias | Description | Mitigation |
|---|---|---|
| Apophenia | Seeing patterns that aren't there | Require multiple independent data points |
| Confirmation bias | Favoring evidence that fits the theory | Actively seek disconfirming evidence |
| Anchoring bias | Over-relying on first information | Revisit initial assumptions regularly |
| Availability heuristic | Overweighting easily accessible info | Diversify sources |
| Dunning-Kruger | Overconfidence in limited knowledge | Identify knowledge gaps explicitly |
| Authority bias | Trusting status over argument | Evaluate arguments on merit |
| Motivated reasoning | Believing what is comforting | Ask: "What would falsify this?" |
Identify these common fallacies in both the official narrative and in the theory being built:
When a fallacy is detected, flag it and restate the argument without the fallacy, or acknowledge the argument collapses.
Build the hypothesis using this structured format:
HYPOTHESIS STATEMENT
─────────────────────────────────────────────
Event: [What happened]
Official narrative: [Summary of official explanation]
Red flags identified: [List from Phase 2]
Primary suspect(s): [From Phase 4, highest MMO score]
Proposed motive(s): [From Phase 3]
Proposed mechanism: [How the act was carried out]
Supporting evidence: [Facts that support the hypothesis]
Predicted consequences: [What else should be true IF this hypothesis is correct]
Falsification criteria: [What evidence would DISPROVE the hypothesis]
Confidence level: [Low / Medium / High] + rationale
When multiple competing hypotheses exist:
Before presenting a theory as credible, apply these five tests:
1. NOTICE → Anomaly or inadequacy in official explanation
2. EXPLORE → Identify candidate actors / known conspiracy patterns
3. DOCUMENT → Gather all relevant facts
4. HYPOTHESIZE → Connect actors, motives, and facts into a coherent theory
5. PREDICT → Derive testable consequences of the hypothesis
6. VALIDATE → Check whether predicted consequences hold; iterate
7. AUDIT → Peer-review for bias and fallacy; refine
When delivering analysis, structure output as:
## Event Analysis: [Event Name]
### Red Flags
- [Flag 1]
- [Flag 2]
### Cui Bono — Beneficiaries
| Actor | Benefit | Plausibility |
|-------|---------|-------------|
| … | … | … |
### MMO Suspect Matrix
| Suspect | Motive | Means | Opportunity | Score |
|---------|--------|-------|-------------|-------|
### Hypothesis
[Structured hypothesis from Phase 8]
### Bias & Fallacy Check
[Any detected biases or fallacies + mitigations]
### Confidence Assessment
[Low/Medium/High + reasoning]
### What Would Change This Assessment
[Falsification criteria]
| Dimension | Questions |
|---|---|
| Who | Who benefits? Who is harmed? Who had means, motive, opportunity? |
| What | What are the strengths/weaknesses of the official narrative? What alternative explanations exist? |
| Where | Where are there real-world analogues? Where can more information be found? |
| When | When did this occur? When does a similar pattern appear historically? |
| Why | Why is this explanation offered? Why might it be incomplete? |
| How | How could the act have been executed? How can the theory be tested? |