Using Cybersecurity MCP Tools
This skill helps you effectively combine MCP tools with the methodology skills in this package.
Discovering Available Tools
Before starting any security task, check which MCP tools you have access to. You can list them or attempt to call them. The key tools and what they unlock:
| If you have... |
You can... |
nmap |
Real port scans, service detection, OS fingerprinting |
nuclei |
Automated vuln scanning against 8000+ templates |
sqlmap |
Actual SQLi testing and exploitation |
ffuf |
Real directory/parameter fuzzing |
shodan |
IP recon, internet device search, CVE lookups by product |
virustotal |
Hash/URL/IP/domain reputation, malware analysis |
mitre-attack |
TTP queries, threat actor profiles, Navigator layers |
radare2 |
Binary disassembly, decompilation, analysis |
yara |
Malware pattern matching on samples |
capa |
Capability detection in executables |
binwalk |
Firmware extraction and analysis |
trivy |
Container and IaC vulnerability scanning |
prowler |
AWS/Azure/GCP compliance auditing |
gitleaks |
Secret scanning in repos |
semgrep |
Static code analysis |
bloodhound |
Active Directory attack path analysis |
searchsploit |
Exploit-DB queries |
maigret |
OSINT username searches across 2500+ sites |
Tool Selection by Task
Recon Phase
shodan → passive IP intel before touching the target
nmap → active port/service enumeration
masscan → fast scan for large IP ranges
whatweb → web tech fingerprinting
waybackurls → historical URL discovery
Vuln Identification Phase
nuclei → broad template-based scanning (run first for quick wins)
ffuf → directory/endpoint discovery
nikto → web server misconfiguration checks
semgrep or gitleaks → if source code is in scope
Exploitation Phase
sqlmap → after manually confirming injection point exists
searchsploit → find public exploits for identified versions
ffuf → parameter fuzzing for logic/auth flaws
Post-Exploitation / AD
bloodhound → map AD attack paths after getting initial access
roadrecon → Azure AD enumeration
Threat Intelligence
virustotal → IOC lookup (hashes, IPs, domains, URLs)
shodan → CVE lookups by product (cves_by_product)
mitre-attack → map findings to ATT&CK techniques
otx → AlienVault OTX for threat feeds
Malware Analysis
yara → pattern matching against known signatures
capa → identify capabilities (network comms, persistence, evasion)
radare2 → deep binary analysis, disassembly
ghidra → decompilation for complex samples
binwalk → firmware or packed files
Cloud Security
trivy → scan container images and IaC templates
prowler → AWS/Azure/GCP compliance checks
Code Security
semgrep → SAST with 5000+ security rules
gitleaks → credentials and secrets in repos
Combining Tools with Skills
Skills provide methodology; MCP tools provide execution. Always load the relevant skill first, then use MCP tools to execute the workflow:
Example: Web Application Pentest
- Load skill:
webapp-pentesting (or api-pentesting for APIs)
- Recon:
shodan → passive intel on target
- Enum:
nmap → port/service scan
- Discovery:
ffuf → directory fuzzing
- Vuln scan:
nuclei → template scan
- Manual testing: per skill methodology
- Map findings:
mitre-attack → tag with ATT&CK techniques
Example: Threat Intel Investigation
- Load skill:
performing-threat-intelligence-analysis (or similar)
- IOC enrichment:
virustotal → hash/domain/IP reputation
- Attribution:
mitre-attack → technique → actor mapping
- Context:
shodan → infrastructure intel on attacker IPs
Example: Malware Sample Analysis
- Load skill:
analyzing-malware-with-dynamic-analysis or reverse-engineering-*
- Initial triage:
yara → known family match
- Capability check:
capa → what does it do?
- Deep analysis:
radare2 or ghidra → disassemble/decompile
- Network IOCs:
virustotal → check C2 domains/IPs
Tool Usage Guidelines
Always verify authorization before active scanning. Tools like nmap, nuclei, and sqlmap actively probe targets and may trigger alerts or cause disruption.
Start passive, go active:
- Passive first:
shodan, waybackurls, virustotal, mitre-attack
- Active second:
nmap, nuclei, ffuf, sqlmap
Confirm before exploiting:
- Never run
sqlmap with --level 5 or --risk 3 without understanding the target
- Never use
hashcat on hashes you didn't capture in the engagement
Tool output is raw data — analyze it:
- A nuclei scan returning 50 findings is not a report. Triage each one.
- A bloodhound graph is not an attack path. Analyze what's actually exploitable.
- A shodan result is not confirmed vulnerable. Verify with targeted testing.
When MCP Tools Are Unavailable
If an MCP tool isn't available, Claude should:
- Provide the exact CLI command the human can run manually
- Explain what to do with the output
- Ask for the output to be pasted back for analysis
Example: "I don't have nmap access, but run this: nmap -sV -sC -p- --min-rate 5000 <target> and paste the output."
MITRE ATT&CK Mapping
Always map significant findings to ATT&CK. Use mitre-attack tools to:
get_technique → look up technique details by ID (e.g., T1059.001)
get_techniques_by_tactic → find all techniques for a tactic
get_group → threat actor profile
get_software → malware/tool profiles
generate_navigator_layer → visual heat map for reports
Include ATT&CK IDs in all findings: T1190 - Exploit Public-Facing Application, T1078 - Valid Accounts, etc.
1---2name: using-cybersecurity-mcp-tools3description: Guides Claude on how to discover, select, and effectively use cybersecurity MCP tools (nmap, nuclei, shodan, virustotal, radare2, bloodhound, trivy, sqlmap, ffuf, MITRE ATT&CK, and more) alongside the skills in this package. Use when performing security assessments, threat intelligence, or any task that benefits from live tool access.4metadata:5 source: https://skillmd.com/skills/njones17/using-cybersecurity-mcp-tools6---78# Using Cybersecurity MCP Tools910This skill helps you effectively combine MCP tools with the methodology skills in this package.1112## Discovering Available Tools1314Before starting any security task, check which MCP tools you have access to. You can list them or attempt to call them. The key tools and what they unlock:1516| If you have... | You can... |17|---|---|18| `nmap` | Real port scans, service detection, OS fingerprinting |19| `nuclei` | Automated vuln scanning against 8000+ templates |20| `sqlmap` | Actual SQLi testing and exploitation |21| `ffuf` | Real directory/parameter fuzzing |22| `shodan` | IP recon, internet device search, CVE lookups by product |23| `virustotal` | Hash/URL/IP/domain reputation, malware analysis |24| `mitre-attack` | TTP queries, threat actor profiles, Navigator layers |25| `radare2` | Binary disassembly, decompilation, analysis |26| `yara` | Malware pattern matching on samples |27| `capa` | Capability detection in executables |28| `binwalk` | Firmware extraction and analysis |29| `trivy` | Container and IaC vulnerability scanning |30| `prowler` | AWS/Azure/GCP compliance auditing |31| `gitleaks` | Secret scanning in repos |32| `semgrep` | Static code analysis |33| `bloodhound` | Active Directory attack path analysis |34| `searchsploit` | Exploit-DB queries |35| `maigret` | OSINT username searches across 2500+ sites |3637## Tool Selection by Task3839### Recon Phase401. `shodan` → passive IP intel before touching the target412. `nmap` → active port/service enumeration423. `masscan` → fast scan for large IP ranges434. `whatweb` → web tech fingerprinting445. `waybackurls` → historical URL discovery4546### Vuln Identification Phase471. `nuclei` → broad template-based scanning (run first for quick wins)482. `ffuf` → directory/endpoint discovery493. `nikto` → web server misconfiguration checks504. `semgrep` or `gitleaks` → if source code is in scope5152### Exploitation Phase531. `sqlmap` → after manually confirming injection point exists542. `searchsploit` → find public exploits for identified versions553. `ffuf` → parameter fuzzing for logic/auth flaws5657### Post-Exploitation / AD581. `bloodhound` → map AD attack paths after getting initial access592. `roadrecon` → Azure AD enumeration6061### Threat Intelligence621. `virustotal` → IOC lookup (hashes, IPs, domains, URLs)632. `shodan` → CVE lookups by product (`cves_by_product`)643. `mitre-attack` → map findings to ATT&CK techniques654. `otx` → AlienVault OTX for threat feeds6667### Malware Analysis681. `yara` → pattern matching against known signatures692. `capa` → identify capabilities (network comms, persistence, evasion)703. `radare2` → deep binary analysis, disassembly714. `ghidra` → decompilation for complex samples725. `binwalk` → firmware or packed files7374### Cloud Security751. `trivy` → scan container images and IaC templates762. `prowler` → AWS/Azure/GCP compliance checks7778### Code Security791. `semgrep` → SAST with 5000+ security rules802. `gitleaks` → credentials and secrets in repos8182## Combining Tools with Skills8384Skills provide methodology; MCP tools provide execution. Always load the relevant skill first, then use MCP tools to execute the workflow:8586**Example: Web Application Pentest**871. Load skill: `webapp-pentesting` (or `api-pentesting` for APIs)882. Recon: `shodan` → passive intel on target893. Enum: `nmap` → port/service scan904. Discovery: `ffuf` → directory fuzzing915. Vuln scan: `nuclei` → template scan926. Manual testing: per skill methodology937. Map findings: `mitre-attack` → tag with ATT&CK techniques9495**Example: Threat Intel Investigation**961. Load skill: `performing-threat-intelligence-analysis` (or similar)972. IOC enrichment: `virustotal` → hash/domain/IP reputation983. Attribution: `mitre-attack` → technique → actor mapping994. Context: `shodan` → infrastructure intel on attacker IPs100101**Example: Malware Sample Analysis**1021. Load skill: `analyzing-malware-with-dynamic-analysis` or `reverse-engineering-*`1032. Initial triage: `yara` → known family match1043. Capability check: `capa` → what does it do?1054. Deep analysis: `radare2` or `ghidra` → disassemble/decompile1065. Network IOCs: `virustotal` → check C2 domains/IPs107108## Tool Usage Guidelines109110**Always verify authorization before active scanning.** Tools like `nmap`, `nuclei`, and `sqlmap` actively probe targets and may trigger alerts or cause disruption.111112**Start passive, go active:**113- Passive first: `shodan`, `waybackurls`, `virustotal`, `mitre-attack`114- Active second: `nmap`, `nuclei`, `ffuf`, `sqlmap`115116**Confirm before exploiting:**117- Never run `sqlmap` with `--level 5` or `--risk 3` without understanding the target118- Never use `hashcat` on hashes you didn't capture in the engagement119120**Tool output is raw data — analyze it:**121- A nuclei scan returning 50 findings is not a report. Triage each one.122- A bloodhound graph is not an attack path. Analyze what's actually exploitable.123- A shodan result is not confirmed vulnerable. Verify with targeted testing.124125## When MCP Tools Are Unavailable126127If an MCP tool isn't available, Claude should:1281. Provide the exact CLI command the human can run manually1292. Explain what to do with the output1303. Ask for the output to be pasted back for analysis131132Example: "I don't have nmap access, but run this: `nmap -sV -sC -p- --min-rate 5000 <target>` and paste the output."133134## MITRE ATT&CK Mapping135136Always map significant findings to ATT&CK. Use `mitre-attack` tools to:137- `get_technique` → look up technique details by ID (e.g., T1059.001)138- `get_techniques_by_tactic` → find all techniques for a tactic139- `get_group` → threat actor profile140- `get_software` → malware/tool profiles141- `generate_navigator_layer` → visual heat map for reports142143Include ATT&CK IDs in all findings: `T1190 - Exploit Public-Facing Application`, `T1078 - Valid Accounts`, etc.