When to use it
Analyze a suspicious message safely without clicking links.
🤖 AI prompt
Phishing Analysis Prompt
By Omellody Editorial Team · Published · UpdatedAnalyze a suspicious message safely without clicking links.
Copy-ready prompt
{message_text}{sender_context}{user_concern}
Analyze this suspicious message safely.
Message: {message_text}
Sender context: {sender_context}
Concern: {user_concern}
Do not click links. Return red flags, legitimacy checks, safe next steps, and a response/no-response recommendation.
Variables
| Variable | How to fill it |
|---|---|
| {message_text} | Replace with your real message text. |
| {sender_context} | Replace with your real sender context. |
| {user_concern} | Replace with your real user concern. |
Quality checks
- Add real context before running the prompt.
- Ask for examples if the first answer is too generic.
- Verify facts, numbers, links, and sensitive advice.
Direct answer
Phishing analysis prompt: Use this prompt to classify a suspicious email or SMS from safe metadata and copied text, then return red flags, verification steps, and a non-clicking response plan.
| Input | What to provide |
|---|---|
| Message type | Email, SMS, chat DM, invoice, login alert, delivery notice. |
| Signals | Sender domain, urgency, attachment type, link text, requested action. |
| Safety | Do not open attachments or visit links; inspect in a safe environment. |
| Output | Ask for risk label, red flags, safe verification path, and escalation wording. |
Source snapshot: Omellody generated this utility page from first-party prompt-library templates and refreshed it during the 2026-05-29 Red Mode quality-rescue pass. The page is not copied from external repositories; external radar signals were used only to understand demand shape.
Reusable output checklist
- Start with a one-sentence direct answer for the requested suspicious message review.
- Return a structured table with priority, reason, draft/output, and verification step.
- Include assumptions and missing information questions only after the first useful draft.
- End with a short safety or quality review so the user can act without exposing sensitive data.
Red Mode depth upgrade: compact scoring rubric
Use this rubric before accepting the AI output for a phishing triage. It turns the prompt into a repeatable mini-tool instead of a one-off draft.
| Score area | Pass condition | Fix if weak |
|---|---|---|
| Completeness | The answer covers sender checks, URL/domain clues, attachment risk, safe reporting steps, and evidence capture. | Ask the model to add missing rows and mark assumptions explicitly. |
| Actionability | Every recommendation has an owner, next step, and verification method. | Request a P0/P1/P2 table with dates or effort levels. |
| Safety | No secrets, credentials, customer data, or private files are requested. | Replace sensitive details with labels such as [internal tool] or [account group]. |
| Reviewability | A human can quickly check facts, risk level, and final decision. | Add a final “review checklist” and “unknowns” section. |
Better follow-up prompts
Improve the draft above for a real phishing triage.
Return:
1. a direct answer,
2. a priority table,
3. assumptions,
4. risks or edge cases,
5. a verification checklist.
Do not request confidential information.
For adjacent workflows, pair this with Password audit prompt or return to the AI Prompt Library.
Related prompt tools
FAQ
How do I use this prompt?
Copy it, replace the variables with real context, and ask follow-up questions to refine the output.
Can I change the structure?
Yes. Add constraints, examples, required format, and quality bar.
Is the output always correct?
No. AI output should be reviewed and verified before publishing or acting on it.