A red team penetration testing agent that automates security assessment tools including network scanning (fscan, httpx, dnsx), vulnerability scanning (nuclei), password attacks (kerbrute), AD enumeration (BloodHound, SharpHound), LLMNR/NBT-NS spoofing (Responder, Inveigh), and lateral movement utilities (impacket, netexec, chisel). Provides payload generation, proxy setup automation, and AD attack path analysis.
RedTeam-Agent is a red-team payload generator with serious definition and composition problems. While 8 tools are declared with descriptions and partial schemas, the definitions are generic, descriptions lack actionable detail for LLM selection, parameter constraints are missing, output schemas are undocumented, and composition violates single-responsibility principle. Tool names are reasonable (verb_noun pattern mostly present), but descriptions are vague about prerequisites, dependencies, and return formats. Many tools combine multiple concerns (e.g., proxy_setup_payload_generator both generates AND provides upload instructions), violating separation of concerns. No evidence of error handling, parameter validation guidance, or idempotent operation semantics. This is a domain-specific payload framework, not a production-grade MCP server.
Generates bind shell scripts that listen on a local port for incoming command execution.
BloodHound JSON data analyzer that loads SharpHound/bloodhound-python collected JSON data and generates comprehensive AD attack path analysis reports including environment overview, high-value targets, session analysis, local admin analysis, group relationships, ACL permissions, domain trusts, GPO analysis, and attack path summaries.
Generates chisel client command for connecting to a chisel server with port forwarding configuration.
Generates chisel server command with optional authentication password for reverse proxy tunneling.
Generates PowerShell reverse shell script that establishes TCP connection and executes remote commands.
Generates complete proxy payload configurations for chisel, netcat, and PowerShell with upload instructions, execute commands, verification commands, and binary information including size, MD5 hash, and base64 encoding for multiple operating systems (Windows/Linux).
Output schemas completely undocumented across all 8 tools. LLMs cannot plan downstream operations or extract relevant fields without knowing the response structure.
Input parameters lack validation constraints (enums, ranges, formats). 'tool' parameters accept free-form strings instead of enums (chisel|nc|netcat|powershell). Ports have no min/max. OS types have no enum. This invites hallucinated values and failed calls.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 37 | 2026-07-28+ | v2 |
Generates comprehensive proxy setup deployment report with binary information, upload instructions, execution commands, and verification steps.
Generates reverse proxy PowerShell or netcat scripts for establishing reverse shells and command execution tunnels.
Violation of single-responsibility principle. proxy_setup_payload_generator combines payload generation, upload instructions, execute commands, verification, binary info, MD5 hashing, and base64 encoding, at least 6 concerns. reverse_proxy_script_generator conflates reverse shells and command tunnels. Should decompose into focused tools (generate_payload, generate_upload_script, generate_verify_script, etc.).
Tool descriptions are generic and lack WHEN-TO-USE guidance. chisel_server_command_generator and chisel_client_command_generator have nearly identical names and purposes, LLM easily conflates them. Descriptions should explicitly distinguish: 'Call chisel_server_command_generator FIRST on your attacker machine to start the listener. Then call chisel_client_command_generator on the target.'
No error handling or recovery guidance. Tools provide no indication of what can go wrong (invalid IP format, port out of range, binary not found, permission denied, etc.) or what to do if a call fails. LLM has no recovery path.
Irreversible operations (reverse shells, bind shells, proxy tunnels) have no dry-run or confirmation step. Agent can accidentally generate production-ready payloads without user approval.
proxy_setup_report_generator accepts 'payload' as an object type with minimal schema. No documentation of required fields, structure, or validation. Silent failures likely if upstream tool output structure changes.