CyberFlair combines technical signals, public data, content analysis, and a supervised AI to produce a score and clear explanations.
Category
We query public security signals (e.g., threat lists) to identify known risks (phishing, malware…). No report ≠ no risk: a site may be new or not yet detected.
Why?
Avoid sites already flagged as dangerous.
How?
Public sources + contextual reading, without overreaching conclusions.
Known threats : spot domains/URLs already reported.
Trust : we display a clear disclaimer about limitations.
Report excerpt (example)
Public data
Two illustrative checks: one reassuring, one to watch.
Score
78
/ 100
Security reports (public source)
No reports detected at the time of analysis. Good indicator, but not an absolute guarantee.
90 / 100
Insufficient public history
Little public information available: not "bad" in itself, but reduces confidence.
45 / 100
Category
Here we look at what the site claims (promises, prices, guarantees, brand) and what it omits (legal mentions, contact, return policy). AI helps spot inconsistencies and suspicious phrasing.
Why?
Scams often betray themselves through text and promises.
How?
Text extraction + rules + AI (coherence, weak signals).
Legal mentions & contact : presence, precision, consistency.
Price consistency : extreme discounts, unrealistic promises.
Report excerpt (example)
Content
What the site says (and forgets to say).
Score
62
/ 100
Language quality
Overall coherent and professional text: fewer "automatic" or contradictory signals.
85 / 100
Legal mentions & contact
Incomplete or hard-to-find information: less transparency = more caution.
35 / 100
Category
Technical signals evaluate the site’s "hygiene": HTTPS security, domain coherence, payment clues, and some infrastructure markers. Useful to avoid basic traps.
Why?
A serious site rarely has an inconsistent technical base.
How?
Technical checks + heuristics (without jumping to conclusions).
HTTPS : secures exchanges (essential, not sufficient).
Domain : structure and deception clues (typos, subdomains, etc.).
Report excerpt (example)
Technical
Technical hygiene + suspicious signals.
Score
74
/ 100
HTTPS security
Valid certificate: communications are encrypted (important if you enter data).
100 / 100
Ambiguous domain/URL
Structure that could cause confusion (brand in a subdomain, variants, etc.).
40 / 100
Category
Performance is not a "scam / not scam" verdict. But an extremely slow, heavy, or overloaded site may reflect a lack of seriousness… or an accumulation of scripts.
Why?
User experience influences trust and credibility.
How?
Load/weight/structure metrics + cautious interpretation.
Speed : perceived/estimated load time.
Weight : heavy page = more friction, sometimes more trackers.
Report excerpt (example)
Performance
A "quality" signal more than a "security" signal.
Score
58
/ 100
DOM complexity
Reasonable structure: the page remains readable and generally more stable performance-wise.
80 / 100
Load speed
Slow loading: increases friction and may degrade the perception of seriousness.
50 / 100
Cross-cutting component
AI does not replace factual signals. It helps to: format the report, detect weak signals (coherence, promises), and produce clear explanations.
What AI does
Coherence, summaries, prioritization, readable explanations.
What AI does not do
Guarantee reliability or "invent" missing information.
Report excerpt (example)
AI analysis (explanations)
AI translates signals into understandable conclusions.
Confidence
0.82
/ 1.00
Overall coherence
The discourse is stable: promises, prices, and conditions seem aligned and understandable.
88 / 100
"Too good to be true" promises
Some marketing elements may push for purchase without sufficient evidence. Caution recommended.
42 / 100
Guidance: if information is not observable, we avoid certainties. We prefer "uncertain" to "invented".
Method
The overall score summarizes category scores. Each check has a contribution and an explanation. Goal: help you decide, not "judge".
Scores per category
Public, Content, Technical, Performance.
Weighting
Some signals count more (e.g., reported threats).
Explained checks
Each point describes a precise signal, not a vague opinion.
Uncertainty
If data is missing, the report can indicate it.
A high score = few negative signals detected. It is not an absolute guarantee.
Transparency
CyberFlair is a decision aid based on signals available at the time of analysis. No tool can guarantee 100%.
New sites
Little or no public traces initially.
HTTPS ≠ honesty
A malicious site can also use HTTPS.
Blocked access
Anti-bot, login, geo-blocking: some pages are inaccessible.
Real context
Seller, offer, terms: everything matters.
FAQ
About the method, the role of AI, and reading a score.
Because the score reflects signals observed on the analyzed URL (content, transparency, performance…), not just popularity.
Yes. That’s why it is supervised, and the report relies primarily on observable signals + explanations.
No. It’s reassuring, but not sufficient. Check the consistency of offers, legal mentions, and terms.
Because some signals are neither good nor bad in isolation: they only make sense when combined.
Multi-signal analysis, clear explanations, supervised AI — and always free.
You can accept or refuse audience measurement and advertising cookies. Necessary cookies (session, security) always remain active. See the Cookies page to refine your choices.
Your choice is stored for approximately 6 months. If the legal texts are updated, we may ask you to choose again.