network-pentest
Fail
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: CRITICALCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATIONPERSISTENCEDYNAMIC_EXECUTIONOBFUSCATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of high-privilege network tools such as
nmap,tcpdump,bettercap,responder, andhping3. These tools can be used to disrupt network services, intercept communications, or perform unauthorized scanning of target environments.\n- [DATA_EXFILTRATION]: The skill provides detailed instructions and payloads for the extraction of sensitive data, including NTLM hashes through Responder poisoning, Kerberos tickets via DCSync and Kerberoasting, and plaintext credentials captured from network traffic through sniffing and man-in-the-middle attacks.\n- [EXTERNAL_DOWNLOADS]: Automated security scans identified references tobook.hacktricks.xyzas malicious (associated with EshopScam). Additionally, the skill's installation guides recommend fetching third-party tools usinggo installandgit clonefrom external GitHub repositories (e.g., chisel, ligolo-ng, assetfinder, subfinder) without integrity verification or version pinning.\n- [PRIVILEGE_ESCALATION]: The skill frequently employssudoto execute tools requiring raw socket access. The documentation also explicitly discusses theoretical and practical privilege escalation techniques as part of the penetration testing workflow.\n- [PERSISTENCE]: The skill documents several persistence mechanisms, including the creation of web shells and the modification of system startup files (cron, shell profiles, registry keys) to maintain access to a compromised system.\n- [DYNAMIC_EXECUTION]: The skill uses dynamic code execution patterns, specificallypython3 -candscapy -cinside the payloads and guides, to execute custom protocol logic and exploits directly from the command line.\n- [OBFUSCATION]: The skill includes instructions for anti-forensics and evasion techniques, such as usingtimestompto alter file timestamps,memfd_create()for fileless execution, and decoy port scans to mask the attacker's true origin.\n- [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by processing untrusted data from network traffic capture and external web resources. Ingestion points are present inguides/comprehensive_pentest_demo.shandpayloads.md. The skill possesses capabilities for raw network access and file writing, but lacks explicit sanitization or boundary markers to prevent the execution of malicious instructions embedded in the processed data.
Recommendations
- AI detected serious security threats
- Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata