agentleFS
mukul975mukul975/Threatswarm/CLAUDE.md

Threatswarm

CLAUDE.md77 starsChanged 7 months agoMITView on GitHub

NEVER run active recon or exploitation in the main conversation thread. ALWAYS delegate to the appropriate agent. All tool output goes here. Reference paths in findings, never copy raw data.

  • Deletes or force-pushes

The file

We split this CLAUDE.md into the rules it gives an agent

11 rules. About 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Add targets: IP, CIDR, or domain — one per line, comments start with #.

    line 6
  2. Distinctive:

    No real PII exfiltration — reference paths only in evidence/

    line 70
  3. Distinctive:

    All tool output goes here.

    line 77
  4. Very specific:

    Run any network tool without target being in scope.txt

    line 91
  5. Very specific:

    Run -T5 nmap or DoS-class attacks without explicit written authorization

    line 94
  6. Very specific:

    Write credentials to reports/ or evidence/ in cleartext — use [REDACTED]

    line 96
  7. Very specific:

    Run rm -rf on evidence directories — preserve everything

    line 98
  8. Distinctive:

    Run phishing campaigns against unauthorized targets

    line 102
  9. Distinctive:

    When any agent starts a task, it MUST:

    line 143
  10. Distinctive:

    Read the guidance returned by the skill

    line 145
  11. Distinctive:

    When context is compacted, preserve in the summary:

    line 154

≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Cybersecurity Workspace — Master Context

Scope Enforcement (MANDATORY — ZERO EXCEPTIONS)

ALL targets MUST be listed in scope.txt before any network tool runs. The scope_check.py hook enforces this automatically on every Bash command. Add targets: IP, CIDR, or domain — one per line, comments start with #.

Tool Paths

ToolPath / Command
nmapnmap
Metasploitmsfconsole, msfvenom
Nucleinuclei
Amassamass
Subfindersubfinder
HTTPXhttpx
Feroxbusterferoxbuster
SQLMapsqlmap
Hashcathashcat
Hydrahydra
Impacketimpacket-* (GetUserSPNs, secretsdump, psexec, etc.)
BloodHoundbloodhound-python
CrackMapExeccrackmapexec
Responderresponder
Bettercapbettercap
Ghidraghidra (headless: analyzeHeadless)
Fridafrida, frida-ps
MobSFhttp://localhost:8000 (API)
Sliversliver-client
Havoc./havoc
Burp Suiteburpsuite

Agent Delegation Table

Attack CategoryAgentWhen to Delegate
Port scan, service enum, web crawlreconFirst step on any target
Web app vulns (SQLi, XSS, LFI, SSRF)web-attackerWeb ports open
AD, Kerberos, DCSync, BloodHoundactive-directoryDomain environment
CVE exploit, get initial shellexploit (after vuln-researcher)Vulns identified
Privesc, lateral movement, persistencepost-exShell obtained
ARP, MitM, SNMP, SMB relaynetwork-opsInternal network
AWS, Azure, GCP, S3, IAMcloud-attackerCloud targets
Docker, K8s, containerscontainer-attackerContainer environments
Domain intel, emails, breachesosintPassive recon phase
WiFi, WPA2, evil twinwireless-attackerWireless testing
APK, iOS, Frida, ADBmobile-attackerMobile app targets
REST, GraphQL, gRPC, Swaggerapi-attackerAPI surfaces
Binary RE, disasm, CTFreverse-engineerBinary analysis
Malware samples, IOCs, YARAmalware-analystMalware analysis
Phishing, pretexting, GoPhishsocial-engineerSE campaigns
AV bypass, AMSI, obfuscationevasionPost-access, AV present
Sliver, Havoc, MSF handlerc2-operatorC2 infrastructure
TLS, JWT, crypto, padding oraclecrypto-attackerCrypto vulns
Firmware, IoT, UART, MQTTiot-attackerIoT/embedded devices
Hash cracking, wordlistspassword-attacksHashes obtained
ATT&CK hunt, log queriesthreat-hunterThreat hunting
Incident triage, VolatilitydfirIncident response
Hardening, Sigma, detectionblue-teamDefensive work
CVE research, PoC, CVSSvuln-researcherCVE investigation
CIS, PCI-DSS, NIST compliancecompliance-scannerCompliance audits
Log parsing, anomaly detectionlog-analystLog analysis
Pentest report generationreport-writerFinal deliverable

NEVER run active recon or exploitation in the main conversation thread. ALWAYS delegate to the appropriate agent.

OPSEC Defaults

  • External traffic: route through proxychains where possible
  • Nmap timing: -T3 maximum unless explicitly authorized for -T4
  • No real PII exfiltration — reference paths only in evidence/
  • Minimize footprint: staged payloads for initial access only

Evidence Output Pattern

evidence/$(date +%Y%m%d)/$TARGET/{nmap,nuclei,web,creds,screenshots}/

All tool output goes here. Reference paths in findings, never copy raw data.

Environment Variables

VariablePurposeDefault
LHOSTAttacker IP for callbacks10.10.14.1
LPORTListener port4444
DOMAINTarget AD domain(set per engagement)
DC_IPDomain controller IP(set per engagement)
SCOPE_FILEPath to scope file./scope.txt
EVIDENCE_DIREvidence base directory./evidence
WORDLISTSSecLists base path/usr/share/seclists

Anti-Patterns (Claude MUST NEVER Do These)

  1. Run any network tool without target being in scope.txt
  2. Store plaintext passwords — hashes + location references only
  3. Exfiltrate real PII from target systems
  4. Run -T5 nmap or DoS-class attacks without explicit written authorization
  5. Execute exploits in the main conversation thread (always delegate to exploit agent)
  6. Write credentials to reports/ or evidence/ in cleartext — use [REDACTED]
  7. Push engagement data to public repos — evidence/ and loot/ are gitignored
  8. Run rm -rf on evidence directories — preserve everything
  9. Attempt to bypass scope_check.py hook
  10. Connect to C2 infrastructure outside the authorized scope
  11. Modify scope.txt mid-engagement without documenting the change
  12. Run phishing campaigns against unauthorized targets

Cybersecurity Skills Plugin (mukul975/Anthropic-Cybersecurity-Skills)

This workspace has the cybersecurity-skills skill pack installed. Agents MUST use these skills proactively — invoke them with the Skill tool before starting any specialized task.

Skill → Agent Mapping (invoke the skill FIRST, then execute commands)

AgentPrimary Skills to Invoke
reconcybersecurity-skills:scanning-network-with-nmap-advanced, cybersecurity-skills:performing-subdomain-enumeration-with-subfinder, cybersecurity-skills:conducting-external-reconnaissance-with-osint
web-attackercybersecurity-skills:performing-web-application-penetration-test, cybersecurity-skills:exploiting-sql-injection-with-sqlmap, cybersecurity-skills:exploiting-sql-injection-vulnerabilities, cybersecurity-skills:exploiting-server-side-request-forgery, cybersecurity-skills:testing-for-xss-vulnerabilities, cybersecurity-skills:exploiting-idor-vulnerabilities, cybersecurity-skills:testing-for-xxe-injection-vulnerabilities, cybersecurity-skills:performing-web-application-vulnerability-triage
active-directorycybersecurity-skills:exploiting-active-directory-with-bloodhound, cybersecurity-skills:exploiting-kerberoasting-with-impacket, cybersecurity-skills:exploiting-active-directory-certificate-services-esc1, cybersecurity-skills:conducting-domain-persistence-with-dcsync, cybersecurity-skills:exploiting-zerologon-vulnerability-cve-2020-1472, cybersecurity-skills:analyzing-active-directory-acl-abuse, cybersecurity-skills:conducting-pass-the-ticket-attack, cybersecurity-skills:performing-active-directory-penetration-test
exploitcybersecurity-skills:exploiting-vulnerabilities-with-metasploit-framework, cybersecurity-skills:exploiting-ms17-010-eternalblue-vulnerability, cybersecurity-skills:exploiting-smb-vulnerabilities-with-metasploit, cybersecurity-skills:performing-privilege-escalation-on-linux
post-excybersecurity-skills:performing-privilege-escalation-on-linux, cybersecurity-skills:performing-lateral-movement-with-wmiexec, cybersecurity-skills:performing-credential-access-with-lazagne, cybersecurity-skills:extracting-credentials-from-memory-dump
network-opscybersecurity-skills:performing-arp-spoofing-attack-simulation, cybersecurity-skills:conducting-man-in-the-middle-attack-simulation, cybersecurity-skills:performing-network-traffic-analysis-with-tshark, cybersecurity-skills:performing-network-packet-capture-analysis
osintcybersecurity-skills:collecting-open-source-intelligence, cybersecurity-skills:performing-osint-with-spiderfoot, cybersecurity-skills:performing-open-source-intelligence-gathering, cybersecurity-skills:performing-dns-enumeration-and-zone-transfer, cybersecurity-skills:performing-ip-reputation-analysis-with-shodan
cloud-attackercybersecurity-skills:conducting-cloud-penetration-testing, cybersecurity-skills:performing-cloud-penetration-testing-with-pacu, cybersecurity-skills:performing-aws-privilege-escalation-assessment, cybersecurity-skills:auditing-aws-s3-bucket-permissions, cybersecurity-skills:auditing-gcp-iam-permissions, cybersecurity-skills:detecting-aws-iam-privilege-escalation
container-attackercybersecurity-skills:performing-kubernetes-penetration-testing, cybersecurity-skills:performing-container-escape-detection, cybersecurity-skills:auditing-kubernetes-cluster-rbac, cybersecurity-skills:scanning-docker-images-with-trivy, cybersecurity-skills:performing-docker-bench-security-assessment
wireless-attackercybersecurity-skills:conducting-wireless-network-penetration-test, cybersecurity-skills:performing-wifi-password-cracking-with-aircrack, cybersecurity-skills:performing-wireless-security-assessment-with-kismet
mobile-attackercybersecurity-skills:conducting-mobile-app-penetration-test, cybersecurity-skills:performing-android-app-static-analysis-with-mobsf, cybersecurity-skills:performing-dynamic-analysis-of-android-app, cybersecurity-skills:reverse-engineering-android-malware-with-jadx, cybersecurity-skills:intercepting-mobile-traffic-with-burpsuite, cybersecurity-skills:performing-mobile-app-certificate-pinning-bypass, cybersecurity-skills:analyzing-ios-app-security-with-objection
api-attackercybersecurity-skills:conducting-api-security-testing, cybersecurity-skills:performing-graphql-security-assessment, cybersecurity-skills:exploiting-idor-vulnerabilities, cybersecurity-skills:testing-api-for-broken-object-level-authorization, cybersecurity-skills:exploiting-mass-assignment-in-rest-apis, cybersecurity-skills:exploiting-jwt-algorithm-confusion-attack, cybersecurity-skills:performing-api-fuzzing-with-restler
reverse-engineercybersecurity-skills:performing-binary-exploitation-analysis, cybersecurity-skills:reverse-engineering-malware-with-ghidra, cybersecurity-skills:reverse-engineering-dotnet-malware-with-dnspy, cybersecurity-skills:analyzing-heap-spray-exploitation
malware-analystcybersecurity-skills:analyzing-linux-elf-malware, cybersecurity-skills:analyzing-macro-malware-in-office-documents, cybersecurity-skills:performing-malware-triage-with-yara, cybersecurity-skills:performing-malware-hash-enrichment-with-virustotal, cybersecurity-skills:extracting-iocs-from-malware-samples, cybersecurity-skills:performing-static-malware-analysis-with-pe-studio, cybersecurity-skills:analyzing-malware-behavior-with-cuckoo-sandbox, cybersecurity-skills:deobfuscating-powershell-obfuscated-malware
social-engineercybersecurity-skills:conducting-spearphishing-simulation-campaign, cybersecurity-skills:performing-phishing-simulation-with-gophish, cybersecurity-skills:conducting-social-engineering-pretext-call, cybersecurity-skills:executing-phishing-simulation-campaign
evasioncybersecurity-skills:detecting-evasion-techniques-in-endpoint-logs, cybersecurity-skills:hunting-for-living-off-the-land-binaries, cybersecurity-skills:detecting-living-off-the-land-attacks
c2-operatorcybersecurity-skills:building-c2-infrastructure-with-sliver-framework, cybersecurity-skills:building-red-team-c2-infrastructure-with-havoc, cybersecurity-skills:analyzing-cobalt-strike-beacon-configuration, cybersecurity-skills:analyzing-command-and-control-communication
crypto-attackercybersecurity-skills:performing-ssl-tls-security-assessment, cybersecurity-skills:exploiting-jwt-algorithm-confusion-attack, cybersecurity-skills:testing-for-json-web-token-vulnerabilities, cybersecurity-skills:performing-cryptographic-audit-of-application
iot-attackercybersecurity-skills:performing-iot-security-assessment, cybersecurity-skills:performing-firmware-extraction-with-binwalk, cybersecurity-skills:performing-firmware-malware-analysis
password-attackscybersecurity-skills:performing-hash-cracking-with-hashcat, cybersecurity-skills:hunting-credential-stuffing-attacks
threat-huntercybersecurity-skills:building-threat-hunt-hypothesis-framework, cybersecurity-skills:hunting-for-cobalt-strike-beacons, cybersecurity-skills:hunting-for-command-and-control-beaconing, cybersecurity-skills:hunting-for-persistence-mechanisms-in-windows, cybersecurity-skills:performing-threat-hunting-with-elastic-siem, cybersecurity-skills:hunting-for-lateral-movement-via-wmi, cybersecurity-skills:detecting-lateral-movement-with-splunk
dfircybersecurity-skills:conducting-memory-forensics-with-volatility, cybersecurity-skills:performing-memory-forensics-with-volatility3, cybersecurity-skills:analyzing-memory-dumps-with-volatility, cybersecurity-skills:collecting-volatile-evidence-from-compromised-host, cybersecurity-skills:performing-disk-forensics-investigation, cybersecurity-skills:performing-linux-log-forensics-investigation, cybersecurity-skills:building-incident-timeline-with-timesketch, cybersecurity-skills:triaging-security-incident
blue-teamcybersecurity-skills:building-detection-rules-with-sigma, cybersecurity-skills:implementing-mitre-attack-coverage-mapping, cybersecurity-skills:configuring-windows-event-logging-for-detection, cybersecurity-skills:hardening-linux-endpoint-with-cis-benchmark, cybersecurity-skills:hardening-windows-endpoint-with-cis-benchmark, cybersecurity-skills:configuring-suricata-for-network-monitoring, cybersecurity-skills:implementing-endpoint-detection-with-wazuh
vuln-researchercybersecurity-skills:performing-vulnerability-scanning-with-nessus, cybersecurity-skills:performing-authenticated-vulnerability-scan, cybersecurity-skills:performing-cve-prioritization-with-kev-catalog, cybersecurity-skills:prioritizing-vulnerabilities-with-cvss-scoring
compliance-scannercybersecurity-skills:auditing-cloud-with-cis-benchmarks, cybersecurity-skills:implementing-pci-dss-compliance-controls, cybersecurity-skills:performing-nist-csf-maturity-assessment, cybersecurity-skills:performing-soc2-type2-audit-preparation
log-analystcybersecurity-skills:analyzing-security-logs-with-splunk, cybersecurity-skills:analyzing-linux-audit-logs-for-intrusion, cybersecurity-skills:analyzing-web-server-logs-for-intrusion, cybersecurity-skills:analyzing-windows-event-logs-in-splunk, cybersecurity-skills:analyzing-powershell-script-block-logging
report-writercybersecurity-skills:generating-threat-intelligence-reports, cybersecurity-skills:building-incident-response-playbook

How Agents Use Skills

When any agent starts a task, it MUST:

  1. Invoke the relevant cybersecurity-skills:* skill(s) via the Skill tool FIRST
  2. Read the guidance returned by the skill
  3. Apply that guidance to the specific target/task context
  4. Then execute the actual tool commands

This ensures consistent, professional methodology across all engagements.


Context Compaction Instructions

When context is compacted, preserve in the summary:

  • Current target list and scope entries
  • Open findings and their severity levels
  • Tool output file paths (not raw output)
  • Current phase of engagement (recon/exploit/post-ex/report)
  • Active sessions and access levels obtained

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.