Fallacy Analyzer — paste an argument, see its fallacies.
A public AI product I built end to end on the Claude API. You submit a short argument and an 8-agent pipeline moderates the input, extracts the argumentative structure, classifies candidate fallacies, runs a verification plan against the web, evaluates each candidate against a typed catalogue of 10 fallacy types, scores the result, and writes a clean Markdown report.
Strict anti-hallucination policy on sources. Public feed of analyses. Hourly window mechanic with a small capacity cap to ration use and create a ritual.
What it does, in one sentence
You submit a short text containing an argument. The product returns a structured analysis: what the argument is, which logical fallacies it commits (or doesn't), why, and which sources back the verification — with a confidence score per detection and a clean human-readable report.
8 agents, one analysis
Moderator
Input filter. Decides allow / block, marks grey-zone content, returns 1-3 topic tags.
Argument parser
Extracts the argumentative structure from the raw text — premises, claim, implicit links.
Fallacy classifier
Proposes candidate fallacies against a typed catalogue of 10 named fallacies, each one defined as a dedicated Skill.
Research planner
Designs verification queries: what to check on the web to confirm or refute each candidate.
Evidence retriever
Runs the queries against the open web and applies a strict source-quality policy: tiers A / B / C / D. Anti-hallucination rule on sources — a claim verified only against weak sources is reported as not verifiable, not as confirmed.
Fallacy evaluator
Invoked once per candidate. Decides whether it is actually a fallacy, with a reasoned justification grounded in the evidence.
Score aggregator
Consolidates per-candidate evaluations into a single typed assessment with confidence scores.
Report writer
Composes the final Markdown report — what the argument is, which fallacies were detected, why, with sources and confidence.
Each agent runs in isolation with its own operating instructions and a strict, typed contract for its output — so the next agent never has to guess what it received. The Claude API is the only model layer.
10 named fallacies, typed
Each fallacy is its own Skill loaded into the classifier and evaluator — definition, prototypical structure, examples and edge cases. The catalogue is extensible without touching agent code.
Hourly window, capped capacity
A small window each hour
Each natural hour opens a small window with a handful of analysis slots. Unused slots are lost; they don't accumulate. Creates expectation and rhythm.
One submission per user per day
Cap prevents flooding and forces selection: which argument do you actually want analysed?
While the pipeline runs, the user sees a step-by-step progress indicator. Every analysis is public by default and lives in the feed alongside the others — so the product is also a small library of argument-quality examples.
Anti-hallucination by source policy
The verification stage doesn't trust any URL the model invents. A typed source policy classifies retrieved evidence into four tiers (A → D) and applies rules:
- Tier A — primary sources, peer-reviewed journals, government data.
- Tier B — established media with editorial process.
- Tier C — secondary blogs, opinion pieces — usable as context, not as proof.
- Tier D — discarded.
A claim verified only against tier C or D is reported as not verifiable rather than confirmed. The policy is defined declaratively as configuration — editable without touching agent code.
What's under the hood
api Pipeline
- Claude API — a combination of different models, each one chosen for the purpose and need of its step
- Multi-agent orchestration
- Live web evidence with source-quality policy
- Strict, typed contracts between agents
- Skill-based fallacy catalogue (extensible)
language Web app
- Public web front, deployed on its own subdomain
- Worker process for the analysis queue
- Public feed with topic tagging and visual hashing
- HTTPS end to end
Want to try it on your own argument?
Free, public, no login needed to read the feed. Submitting an analysis only requires a Google sign-in.
arrow_outward Open falacias.cordcozy.com