Repository navigation
OpenClaw Use Cases and Real World Examples
A practical guide to getting value out of OpenClaw across personal, professional, IT support, and cybersecurity workflows. These examples assume OpenClaw is running with Telegram or Discord configured as a chat channel so you can interact from anywhere.
Copy-paste-ready example prompts organized by workflow domain (personal, business, IT field work, blue/red/purple team, OSINT), assuming an already-running OpenClaw instance (see README.md for setup). Complements agent_skill_config.md: this file shows what to ask, that file shows how to configure the agent that answers well.
Ten sections of example prompts organized by domain, plus a model-routing strategy table (which model for which task type) and a tips section on prompt specificity, role-priming, and keeping sensitive data local via @ollama.
Have ready-made, tested prompt patterns for common IT/security workflows so you're not starting from a blank chat box for routine tasks.
- Looking for a starting-point prompt for a specific workflow (IT field diagnostics, blue team log triage, OSINT recon, etc.)
- Deciding which model to route a task to (Section: Model Routing Strategy)
- Handling sensitive client data and needing the local-only
@ollamapattern
- Day to Day Personal Life
- Work & Business Productivity
- IT Support & Field Work
- IT Support Business Operations
- Homelab Management
- Blue Team / Defensive Security
- Red Team / Offensive Security
- Purple Team Operations
- OSINT & Reconnaissance
- Model Routing Strategy
- Tips for Getting the Best Results
OpenClaw connected to Telegram means your AI assistant is always in your pocket. These are tasks you'd normally have to sit down at a computer for.
"Give me a quick summary of what's on my calendar today, any unread emails that look urgent,
and today's weather in Vancouver WA"
Requires: Google Calendar skill, Gmail skill, web search
"I'm at a client site and they mentioned moving to Azure AD.
Give me a 5-minute briefing on what that means for a 15-person SMB
migrating from on-prem AD - costs, gotchas, and what questions I should ask them"
"I need to pick up parts for a network install this afternoon.
Make me a checklist: 8-port managed switch, patch cables cat6,
keystone jacks, patch panel, rack screws, and velcro ties.
Estimate cost from Best Buy vs Newegg vs Amazon"
"Explain VLAN trunking to me like I'm prepping for a CompTIA Network+ exam.
Use an analogy, then give me 5 practice questions."
"I have 30 minutes and a set of dumbbells. Give me a quick strength circuit
I can do between service calls. No jumping, I'm in dress clothes."
"Check my Home Assistant - is the garage door open?
Also what's the current temperature inside vs outside?"
Requires: Home Assistant skill or API integration
"Draft a professional follow-up email to a client who hasn't responded
to my quote for a network upgrade in 2 weeks. Keep it friendly but nudge them.
My business is Pacific Northwest Computers."
"I have a discovery call with a new small business client in 20 minutes.
They're a 12-person dental office. Give me the top 10 questions I should ask
to scope their IT needs, and common issues dental offices have with IT compliance."
"I'm uploading this 40-page vendor contract.
Summarize the key terms, flag anything unusual,
and highlight any auto-renewal clauses or liability caps."
"Create a service quote template for a small business network installation.
Include sections for: labor, hardware (with markup), travel, and ongoing support.
Format it so I can fill in the blanks quickly on-site."
"I just fixed a weird issue where a client's Windows 11 machine
kept dropping off the domain after sleep. The fix was disabling
fast startup and setting the NIC to never sleep in power management.
Write this up as a KB article I can add to my GitHub repo."
This is where OpenClaw shines for solo technicians. You're on-site, hands full, and need answers fast via Telegram.
"Client's printer keeps showing offline even though it's on and connected.
It's an HP LaserJet Pro M404 on a Windows 10 machine.
Walk me through a quick diagnostic starting with the most likely causes."
"Client reports 'slow internet' but speedtest shows 150/20 which matches their plan.
They say video calls keep dropping. What are the likely causes and
how do I diagnose this efficiently on-site in under 20 minutes?"
"Give me the PowerShell one-liner to check if Windows Defender
real-time protection is enabled on a remote machine"
"What's the command to flush and re-register DNS on Windows 11?
Also give me the ipconfig release/renew sequence."
"Client has a Dell OptiPlex 7090 running Windows 11 23H2.
They bought a new Logitech C920 webcam and it's not being detected.
What should I check first?"
"Write a non-technical explanation I can send to a client explaining
why their computer is running slow. The actual cause is a failing HDD
with bad sectors. Don't use jargon - they're a 65-year-old small business owner."
"A client wants to know the difference between their router,
modem, and the switch I just installed. Give me a simple analogy
I can use to explain it in under 2 minutes."
"I'm about to remote into a Windows Server 2019 machine that
a client says 'just stopped working.' Give me a first-5-minutes checklist
of what to check before I start making changes."
"I have 4 service calls today. Help me route them efficiently:
- 9am: Battle Ground (PC repair, ~1hr)
- Vancouver downtown (network install, ~3hr)
- Camas (virus removal, ~1.5hr)
- Vancouver near 78th (printer setup, ~30min)
What order makes the most sense geographically?"
"I have 8 open tickets. Here's my list: [paste ticket list].
Which ones are at risk of breaching a 24hr response SLA
based on when they were opened? Flag any over 20 hours."
"Create a client onboarding checklist for a new small business
managed services client. Include: network documentation,
software inventory, backup verification, security assessment,
and a 90-day check-in schedule."
"A client needs:
- Replace 3 workstations (mid-range office use)
- New 8-port managed switch
- NAS for file sharing with 10TB usable
- Setup and configuration labor
Give me a rough quote range with parts at retail and
suggest what markup is typical for MSPs in 2026."
"I just finished a 3-hour network install at ABC Dental.
I replaced their unmanaged switch with a UniFi USW-Lite-8-PoE,
ran 4 new ethernet drops, set up VLANs for guest wifi and office,
and updated firmware on their router.
Write a professional service completion report I can email to the client."
"Write 3 social media posts for my IT business (Pacific Northwest Computers)
promoting our remote support services.
Target small business owners in Vancouver WA.
Keep them under 280 characters and avoid tech jargon."
"Check my TrueNAS pools - are they healthy?
Any scrub errors or degraded drives?"
Requires: TrueNAS skill from ClawHub
"What apps are currently running on my TrueNAS?
Which ones have updates available?"
"List all running VMs on my Proxmox cluster and their resource usage.
Flag anything using over 80% CPU or RAM."
"Any new unknown devices on my network in the last 24 hours?
Check NetAlertX for me."
"When did my last TrueNAS replication task run and was it successful?
Also check if any ZFS snapshots are older than 7 days and should be pruned."
"Quick health check - are all my self-hosted services responding?
Check: Jellyfin, Frigate, Home Assistant, NPMplus, Ollama"
OpenClaw with appropriate skills becomes a hands-free SOC assistant.
"I'm going to paste some Wazuh alerts from the last hour.
Triage them by severity, identify any that look like real threats vs noise,
and suggest immediate actions for anything critical."
"Here are 50 lines of auth.log from my SSH server.
Are there any signs of credential stuffing or brute force?
Give me the source IPs and recommended firewall rules."
"Write a Wazuh rule to detect when a new user account is created
on a Windows machine outside of business hours (8am-6pm PST)"
"Give me Zeek/Suricata rules to detect DNS tunneling based on
high-frequency queries, long subdomain strings, and unusual record types"
"A Windows workstation on my network made 3,000 DNS queries in 5 minutes
to random subdomains of a .xyz domain. Walk me through an IR playbook
for suspected C2 beaconing - what do I contain, collect, and analyze first?"
"I found this suspicious PowerShell command in my logs: [paste command].
Decode it, explain what it does, assess the risk level,
and tell me what artifacts to look for on the host."
"Here's my Nessus scan output for my home network: [paste results].
Prioritize the findings by exploitability and suggest a patching order.
Flag anything with a public exploit."
"Write a 5-minute phishing awareness training script
I can use with a client's non-technical staff.
Include 3 real-world examples and a quiz at the end."
"I'm getting too many false positives from this Wazuh rule: [paste rule].
Suggest how to tune it to reduce noise while keeping coverage
for actual brute force attempts."
For authorized penetration testing engagements. Always ensure proper written authorization before any testing.
โ ๏ธ These examples are for authorized penetration testing only. Ensure you have written permission before testing any system.
"I'm starting a pentest engagement on a Windows Active Directory environment.
Give me a methodology checklist from initial access through domain compromise,
with the key tools at each stage."
"What are the current top techniques for AD enumeration without triggering
common EDR solutions? Focus on living-off-the-land approaches."
"Explain how Kerberoasting works, why it's effective against service accounts
with weak passwords, and what defenders typically see in logs when it's performed."
"I have a shell on a Windows 10 machine as a standard user.
List the top 5 local privilege escalation techniques I should check,
ordered by likelihood of success in a corporate environment."
"I found an unauthenticated RCE vulnerability in a client's web application
via an unsanitized file upload endpoint. Write this up as a professional
pentest finding including: severity rating, technical description,
proof of concept summary, business impact, and remediation steps."
"Convert these raw pentest notes into an executive summary
suitable for a non-technical C-suite audience: [paste notes].
Lead with business risk, not technical details."
"Give me the Nmap command to do a full TCP scan with service detection
and default scripts on a /24 subnet, output to all formats,
without being too aggressive on timing"
"What's the Metasploit module for EternalBlue and what are
the exact prerequisites for it to work?
What should I check before running it?"
"I'm stuck on a CTF challenge. Here's what I have so far: [describe challenge].
Don't give me the answer - give me hints and point me toward the right technique."
Purple team is about blue and red working together to improve defenses. OpenClaw helps coordinate and document this work.
"I want to test whether my Wazuh SIEM detects a Pass-the-Hash attack.
Give me: (1) the exact attack steps to simulate it,
(2) what log events should fire,
(3) what a Wazuh rule to detect it should look like,
and (4) how to verify detection worked"
"Using Atomic Red Team, which atomics should I run to test detection
coverage for T1003 (OS Credential Dumping) on Windows?
Give me the commands and what I should see in Wazuh/Sysmon if detection is working."
"Here are the MITRE ATT&CK techniques from my last red team engagement:
[paste technique list].
Map these against my current detection stack (Wazuh + Zeek + Sysmon)
and identify which techniques I have no detection coverage for."
"Facilitate a tabletop exercise for a ransomware scenario targeting
a 20-person dental office. Give me: the scenario narrative,
inject questions for each phase (initial access, lateral movement,
encryption, extortion), and what decisions the team needs to make at each stage."
"I claimed in a security assessment that our EDR would detect
Mimikatz credential dumping. Design a test to validate this claim
that I can run in a lab environment safely."
For legitimate research, due diligence, and authorized investigations.
โ ๏ธ OSINT techniques should only be used for lawful purposes. Always comply with applicable laws and platform terms of service.
"I'm meeting with a potential client - ABC Manufacturing in Vancouver WA.
Find what you can about their business: size, industry, any news,
tech stack hints from their website, LinkedIn presence,
and any public information about their IT infrastructure."
"Do passive recon on example.com.
Find: DNS records, mail servers, any subdomains indexed publicly,
associated IP ranges, hosting provider, tech stack from headers,
and any data breach mentions in public sources."
"For an authorized background check on a potential employee,
find publicly available information about [name] in [city].
Look for: LinkedIn, professional credentials, any public records,
and verify their claimed certifications are real."
"Summarize what is publicly known about the threat actor group
tracked as 'BlackCat/ALPHV'. Include: TTPs, typical targets,
known IOCs, and recent activity."
"Search HaveIBeenPwned and other public sources for any data breaches
associated with the domain pnwcomputers.com.
What credentials or data may be exposed?"
"I'm doing a security awareness session for a client's staff.
Using only publicly available information about their company,
show me what a social engineer could find out in 15 minutes
to craft a convincing pretexting call.
This is for defensive awareness - show them what attackers can learn."
"Set up a summary: are there any lookalike domains registered
similar to pnwcomputers.com that could be used for phishing?
Check for common typosquatting patterns."
Knowing which model to use for which task saves money and gets better results.
| Task Type | Recommended Model | Reason |
|---|---|---|
| Quick lookups, commands, checklists | Groq / Llama 3.3 70B | Fast, free, good enough |
| Casual chat, scheduling, reminders | Groq / Llama 3.3 70B | Free tier handles this well |
| Long document analysis | Gemini 3.1 Pro | 1M token context window |
| Web research summaries | Gemini 3.1 Pro | Built-in Google Search integration |
| Complex security analysis | Claude Opus | Best reasoning for nuanced tasks |
| Report writing, professional docs | Claude Haiku or Sonnet | Good quality, lower cost than Opus |
| Sensitive/private tasks | Ollama (local) | Model inference can remain local, but connected tools, web lookups, telemetry and agent actions may still send data externally; verify configuration and egress |
| Offline/air-gapped work | Ollama (local) | No internet required |
In the OpenClaw dashboard:
- Set Adaptive mode to let OpenClaw auto-route based on task complexity
- Or prefix messages with the model name:
@groq what's the ping command for Linux - Use
@opuswhen you need Claude's best reasoning on a hard problem
Instead of: "Help me fix this network issue"
Try: "Client has a Ubiquiti UniFi setup, USG-3P router, 2 APs, 8-port switch. Clients on VLAN 10 can't reach the internet but the gateway ping works. What should I check?"
"Act as a senior penetration tester reviewing my methodology.
Here's my recon phase notes for an engagement: [notes].
What am I missing and what would you prioritize next?"
"Act as a non-technical small business owner.
Review this security assessment report I wrote and tell me
what's confusing, what's missing, and what would concern you most."
"1. Find the CVE for the recent Fortinet SSL VPN vulnerability from early 2026
2. Explain the technical details in plain English
3. Write a client advisory email warning them to patch immediately
4. Give me the PowerShell or CLI command to check if their firmware is vulnerable"
"I'm about to run this Bash script on a client's production server.
Review it for anything dangerous, unintended side effects,
or edge cases I should know about before running it: [paste script]"
These mechanisms are not interchangeable. Use a reusable skill when you want consistent triage instructions; use cron/scheduled jobs for fixed-time work; use an event hook when a supported session/tool event fires; use an authenticated webhook/integration for an external Wazuh alert.
| Example | Recommended Mechanism | Verification & Boundary |
|---|---|---|
| Daily security digest at 8:00 AM | Scheduled job / cron | Check timezone, source freshness, actual run and destination |
| Weekly summary of open service tickets | Scheduled job reading permitted records | Check data access, filter out stale/closed tickets, verify coverage |
| Wazuh critical alert from another service | Authenticated integration/webhook, then bounded triage | Validate sender, event ID, duplicate handling and severity |
| Save a summary before an agent lifecycle event | Supported internal hook with reviewed handler | Test actual event behavior; restrict stored data and permissions |
| Repeated vulnerability-triage checklist | Reusable skill or prompt template | Test with synthetic positive/negative samples |
Setup order: (1) define a bounded task and authorized sources, (2) select the correct triggering mechanism, (3) validate input/authentication and privacy, (4) test dry-run and failure cases, (5) inspect the actual result and record it, (6) only then enable it within the intended environment.
A hook firing or scheduled job running does not prove that the final report is correct. Track the deliverable with measurable acceptance criteria and evidence. See Notes, Memory, Hooks & Automation, Milestones & Acceptance Criteria and OpenClaw Hooks & Triggers.
For anything involving:
- Client PII or confidential data
- Internal network diagrams or credentials
- Security assessments with real vulnerability details
...always use @ollama to keep it local:
"@ollama Review this network diagram and identify security gaps: [diagram details]"
# Explain something simply
"Explain [technical concept] to a non-technical client in under 3 sentences"
# Generate a checklist
"Give me a pre-flight checklist before [task] on a production system"
# Write a ticket/report
"Write a service ticket for: [description of issue and resolution]"
# Command lookup
"What's the [Windows/Linux/macOS] command to [task]? Give me the exact syntax."
# Script review
"Review this script for security issues, bugs, or unintended behavior: [script]"
# Incident response
"Walk me through IR steps for: [incident type] on [system type]"
# Client communication
"Write a [friendly/formal/urgent] email to a client explaining: [situation]"
# Cost estimate
"Rough cost estimate for: [hardware/service list] - parts + typical labor"
# Study/cert prep
"Give me 10 practice questions for [certification] on the topic of [topic]"
# Log analysis
"Analyze these logs and tell me if anything looks suspicious: [logs]"
Guide written for use with OpenClaw 2026.4.15 deployed on TrueNAS SCALE. All security examples assume authorized testing environments.
- README.md - OpenClaw setup (required before use cases apply)
- agent_skill_config.md - Agent personas referenced in these examples
- ../ai_prompts.md - General prompt engineering patterns behind these examples
- ../README.md - AI section index
๐ Source: AI/OpenClaw/use_cases.md | โ๏ธ Suggest an edit | โฌ
๏ธ Back to Artificial Intelligence
- ๐บ๏ธ Browse by Topic
- ๐ All Repository Files
- ๐งญ Learning Paths
- ๐ Master Guide
- ๐ค Glossary
- โ๏ธ Legal & Disclaimer
๐ Foundations & Learning
๐ก๏ธ Defense & Investigations
- Incident Response
- Open Source Intelligence (OSINT)
- Reverse Engineering
- Operations Security (OPSEC)
- Compliance
โ๏ธ Platforms & Infrastructure
๐ก Hardware, Wireless & Field Projects
- PNWC Cyberdeck
- Hardware Testing
- Hardware Hacking
- Software-Defined Radio
- Wardriving
- uConsole
- Space Security
๐งฐ Assessment & Field Reference
๐ฆ Downloads & Project Resources
๐๏ธ Additional Topics