skills/mukul975/anthropic-cybersecurity-skills/exploiting-smb-vulnerabilities-with-metasploit/Gen Agent Trust Hub
exploiting-smb-vulnerabilities-with-metasploit
Warn
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: Provides shell commands for network scanning, service enumeration, and exploitation tools such as Metasploit (msfconsole), CrackMapExec, and Responder.
- [REMOTE_CODE_EXECUTION]: Includes instructions for utilizing Metasploit modules to exploit critical vulnerabilities like EternalBlue (MS17-010) and perform lateral movement via PsExec.
- [DATA_EXFILTRATION]: Contains procedures for enumerating network shares, listing files, and extracting password hashes from compromised hosts using meterpreter commands like hashdump.
- [CREDENTIALS_UNSAFE]: References hardcoded example NTLM hashes and credentials within the documentation for demonstration purposes (e.g., in Step 3 and Step 5 of SKILL.md).
- [PROMPT_INJECTION]: The scripts/agent.py script creates a surface for indirect prompt injection by ingesting unverified data from remote targets.
- Ingestion points: smb.listShares() and smb.getServerOS() in scripts/agent.py.
- Boundary markers: Absent. The script outputs raw data to a JSON report.
- Capability inventory: Network communication via Impacket and file-writing via json.dump().
- Sanitization: Absent. The script performs minimal string slicing to remove null terminators but no security-focused validation or escaping.
- [EXTERNAL_DOWNLOADS]: Directs users to install external dependencies like the impacket library from public package registries.
Audit Metadata