Loading ToolBench
MCP servers for HackCode security tools - exploitation, reconnaissance, web testing, password attacks, and utilities
This server exposes 10 security/exploitation tools with significant definition gaps. All tools have descriptions (10-50 words), but most lack proper parameter validation guidance, output schemas are undocumented, and error handling is absent. Naming is direct (msfconsole, crackmapexec) but generic descriptions don't explain WHEN to use each tool vs alternatives. No tool annotations (destructiveHint/readOnlyHint) despite 6 tools marked DESTRUCTIVE. Parameters like 'flags' are free-form strings inviting injection. No evidence of input validation, rate limiting, or permission gates for dangerous operations. Schemas are present but minimal, most lack descriptions for optional parameters (encoder, output, format defaults). Output structure is completely undocumented; LLMs cannot plan downstream calls or extract results reliably.
Custom wordlist generator that spiders a website and extracts words for targeted password attacks.
Network exploitation swiss army knife. Test creds, execute commands, dump SAM, spray passwords across SMB/WinRM/SSH/LDAP.
GPU-accelerated password cracker. Fastest hash cracking for large wordlists.
Identify hash algorithm type. Useful before cracking to determine the correct mode.
Online password brute force tool. Supports SSH, FTP, HTTP, SMB, RDP, MySQL, and 50+ protocols.
Impacket toolkit - psexec, wmiexec, smbexec, secretsdump, getTGT, etc. for Windows/AD exploitation.
John the Ripper password cracker. Cracks hashes from /etc/shadow, SAM, NTDS.dit, and more.
No output schemas documented. LLMs cannot predict what fields are returned, breaking tool chaining and result extraction. E.g., does msfconsole return structured JSON or raw text? Does crackmapexec return per-host success/failure or a summary?
Free-form 'flags' parameters (msfconsole, crackmapexec, impacket, responder, hydra, john, hashcat, cewl) invite command injection and hallucinated values. No validation guidance or enum constraints. LLMs will pass arbitrary strings.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | 2025-06-18+ | v2 |
Run Metasploit Framework commands. Search for exploits, configure and run modules.
Generate payloads with Metasploit. Create reverse shells, meterpreter payloads, and encoded shellcode.
LLMNR/NBT-NS/mDNS poisoner for capturing NTLMv2 hashes on local networks.
No tool annotations (destructiveHint, readOnlyHint, idempotentHint). Six tools are marked DESTRUCTIVE (msfconsole, msfvenom, crackmapexec, impacket, hydra, responder) but LLMs have no machine-readable signal to avoid accidental execution or require confirmation.
No error handling guidance. Descriptions lack recovery hints (e.g., 'If target unreachable, verify network connectivity'). LLMs cannot distinguish retryable vs fatal errors or know what to do next.
Parameter descriptions lack format/constraint guidance. E.g., 'lport' (hydra, msfvenom) has no range (1 - 65535). 'wordlist' paths have no validation. 'target' CIDR notation is undocumented. LLMs will pass invalid values.