01
🎯 Core principles
| Principle | Description | Example |
|---|---|---|
| Context first | Give background (role, goal, constraints) before the question. | "You are a Linux security expert. Analyse the following n8n workflow for vulnerabilities: [content]" |
| Roles & tone | Define your role and the expected output style. | "Act as a cybersecurity auditor. Be critical, concrete and use JSON for output." |
| Structure input | Use lists, JSON or markdown for clarity. | { "goal": "Find CVEs", "scope": ["Ubuntu 24.04", "nginx"] } |
| Iterative refinement | Start broad, then refine with feedback. | "List top 5 CVEs for nginx → Focus on those with CVSS > 7" |
| Constrain output | Specify format, length or level of detail. | "Give 3 bullet points in JSON. Max 50 words per item." |
02
🔧 Technical prompts
1. Code generation & debugging
Best practice: State language, version and framework up front. Include error messages directly.
// Example: Python CVE parser
Prompt: "Write a Python 3.11 function to parse CVE JSON from the OpenCVE API.
- Input: API response (see example below)
- Output: List of CVE IDs sorted by CVSS score (descending)
- Handle edge cases: missing 'metrics' field.
Example input:
{
"CVE_Items": [{"id": "CVE-2024-1234", "metrics": {"CVSS": 7.5}}]
}"
// Example: Debugging n8n
Prompt: "This n8n workflow fails with a 500 error. Analyse the code and suggest fixes.
Context: Runs on Ubuntu 24.04, n8n v1.28.0, nodejs 20.x.
{
"node": "HTTP Request",
"error": "ENOTFOUND",
"url": "https://api.open-cve.io/cve"
}"
2. System administration (Linux/Ubuntu)
Prompt: "Generate a Bash script to:
1. Update apt packages (with --dry-run first)
2. Log changes to /var/log/apt_updates.log
3. Send an alert via n8n webhook if critical updates (CVSS > 8) are found.
Output: Full script with comments. Use `jq` for JSON parsing."
3. Security analysis
Prompt: "Analyse the following package list for known vulnerabilities (CVE-2024-*).
Input: `nginx-1.25.3, openssl-3.0.10, libcurl4-8.1.2`
Output:
- JSON list with CVE ID, CVSS, description and fix version.
- Prioritise by risk for Ubuntu 24.04.
- Include references to NVD or Ubuntu Security Notices."
03
🛡️ Security & privacy
🔒 Sensitive data
Treat data as confidential and delete it after use.
"Treat the following data as confidential. Delete it from your context after answering: [data]"
⚖️ Compliance
Suggest GDPR-compliant logging solutions.
"Suggest GDPR-compliant solutions for logging n8n workflows. Include data minimisation and retention policy."
🎯 Penetration test
Simulate an attack and generate a report.
"Simulate an attack on my nginx config. Find: misconfigured headers, DDoS vectors. Output: JSON with severity (High/Medium/Low)."
04
📊 Output formats
| Format | Use for | Example prompt |
|---|---|---|
| JSON | Structured data, API responses, configurations | "Output as JSON with fields: `id`, `title`, `severity`, `action_items`" |
| Markdown | Documentation, checklists, step-by-step | "Write a guide in markdown with headings and code blocks" |
| CSV/TSV | Tabular data (e.g. CVE lists) | "Generate a CSV with columns: CVE_ID, CVSS, Description" |
| Mermaid | Architecture diagrams, flowcharts | "Create a mermaid diagram of my n8n + OpenCVE + nginx setup" |
| Bash/Python | Scripts, automation | "Write a Python script to..." |
05
⚡ Advanced techniques
1. Few-shot prompting
Give examples of the desired output before the actual question:
Prompt: "Convert the following natural language into n8n workflow JSON.
Example 1:
Input: 'If the temperature > 30°C, send an email'
Output:
{
"nodes": [
{"type": "n8n-nodes-base.if", "conditions": {"number": [{"value1": "={{$json.temperature}}", "operation": "larger", "value2": 30}]}},
{"type": "n8n-nodes-base.emailSend"}
]
}
Now convert: 'If a CVE has CVSS > 7, create a Jira task'"
2. Chain-of-Thought (CoT)
Ask for a step-by-step explanation:
Prompt: "Explain how you would debug a slow n8n workflow.
- Start by listing possible causes (CPU, DB, external APIs).
- For each cause, describe how you test it.
- Output: Numbered list with sub-points."
3. Temperature & creativity
🌡️ Low temperature (0.2-0.5)
Factual tasks (code, data analysis).
🔥 High temperature (0.7-1.0)
Brainstorming, creative solutions.
Prompt: "Suggest 10 creative ways to use n8n + Mistral AI for cybersecurity. // Set temperature to 0.9"
06
🚫 Avoid these mistakes
❌ Too vague a question
Wrong: "Help me with n8n"
Fix: "How do I configure OAuth2 in n8n with Keycloak?"
❌ No context
Wrong: "Fix my code"
Fix: "Here is my Python code (see below). It fails with a `TypeError`. Why?"
❌ Prompt too long
Wrong: Writing a novel in the prompt.
Fix: Split into sub-questions. Use references.
❌ Ignoring constraints
Wrong: "Write a full n8n server"
Fix: "Write a *module* to handle CVE alerts in n8n"
❌ Unclear format
Wrong: "Give me data"
Fix: "Give the data as JSON with fields X, Y, Z"
07
🔍 Domain-specific examples
🔒 Cybersecurity
Prompt: "I found this open port on my server: 10.4.117.25:8080.
- Identify possible services (e.g. n8n, Jenkins).
- Suggest hardening measures.
- Output: JSON with `service`, `risk`, `mitigation_steps`"
🤖 n8n automation
Prompt: "Design an n8n workflow that:
1. Fetches new CVEs from the OpenCVE API every 6 hours.
2. Filters for CVSS > 7 and relevant to Ubuntu 24.04.
3. Sends a Slack message to #security-alerts.
4. Creates a Jira task for each critical CVE.
Output:
- Workflow JSON
- Required n8n nodes and configurations
- Error handling (e.g. API down)"
🖥️ Proxmox/AdGuard
Prompt: "How can I log and analyse DNS queries from my AdGuard server (10.4.117.4)?
- Goal: Detect suspicious activity (e.g. C2 domains).
- Output:
- Bash script to parse AdGuard logs
- Rules to block known malicious domains
- Integration with n8n for alerts"
08
📌 Quick reference: prompt templates
| Use case | Template |
|---|---|
| Code review | "Analyse this [language] script for security holes. Focus: [XSS/SQLi/buffer overflow]. Output: JSON with `line`, `issue`, `severity`, `fix`." |
| Documentation | "Write a README.md for my n8n CVE alert system. Include: setup, dependencies, examples." |
| Troubleshooting | "My [service] returns [error]. Here are the logs: [content]. Explain the cause and suggest fixes." |
| Automation | "Automate the following manual process: [description]. Use n8n + [APIs]. Output: Workflow JSON." |
| Training/data | "Generate a synthetic dataset with [X] rows for [purpose]. Format: CSV. Include columns: [list]." |
09
🛠 Tools & integrations
🤖 n8n + Mistral
Use the HTTP Request node to call the Mistral API with structured prompts.
🔮 Clairvoyance AI
Integrate with n8n to analyse logs or generate reports.
🐳 Ollama
Run models locally for sensitive tasks (e.g. CVE analysis).
10