s0ld13rr/pentestcode

PentestCode - Multi-agent AI penetration testing system with persistent engagement state, strategic coordination, and parallel autonomous operations.

What it solves

PentestCode automates the complex, repetitive process of penetration testing. It replaces the manual cycle of running security tools, parsing their output, and deciding the next step with an autonomous agent that can execute full attack chains—from initial scanning to gaining domain administrator privileges—while maintaining a structured record of all findings.

How it works

The system uses a strategist-coordinator architecture consisting of 13 specialized agents. A lead agent plans the engagement and dispatches tasks to sub-agents (such as recon, scanner, exploiter, and identity agents) who work in parallel.

These agents share a centralized "engagement state," a structured memory that tracks hosts, services, vulnerabilities, credentials, and network relationships. This state allows the agent to perform complex operations like cost-based attack path finding (using Dijkstra and Yen's algorithms) and ensures that findings from one agent are immediately available to others. It integrates with 20+ LLM providers and uses dedicated parser tools to ensure tool outputs (like nmap or sqlmap) are correctly ingested into the shared state.

Who it’s for

  • Penetration Testers: For automating full methodologies from recon to reporting.
  • CTF Players: For flag tracking and multi-target coordination.
  • Bug Bounty Hunters: For automating web app and API security recon.
  • Security Researchers: For systematic enumeration and vulnerability validation.

Highlights

  • Multi-Agent Architecture: Uses a lead strategist and 12 specialist agents for parallel execution.
  • Shared Engagement State: Maintains a persistent, queryable database of the network topology, credentials, and attack paths.
  • Extensive Tool Integration: Includes 18 built-in tools and specialized parsers for common security utilities like nmap, nuclei, and bloodhound.
  • Flexible Execution Modes: Offers auto (autonomous), free (direct requests), and guided (step-by-step approval) modes.
  • Knowledge Packs: Uses on-demand "skills" (markdown files) to provide domain-specific knowledge without bloating the LLM context.

Related

  • Project
    GH05TCREW/pentestagent

    An AI-powered penetration testing framework that automates security assessments using autonomous agents, multi-agent orchestration, and MCP-integrated security tools.

  • Project
    SHAdd0WTAka/Zen-Ai-Pentest

    An autonomous, AI-driven penetration testing framework that orchestrates over 72 security tools to automate reconnaissance, exploitation, and reporting.

  • Project
    Autumn-27/ARTEX

    An AI-driven autonomous penetration testing system that uses a multi-agent architecture and dual-graph tracking to automate security auditing and attack chain execution.

  • Project
    vxcontrol/pentagi

    An autonomous AI-powered penetration testing platform that uses a multi-agent system to automate security vulnerability research and exploitation in isolated environments.

  • Project
    bugbasesecurity/pentest-copilot

    An AI-driven penetration testing agent that autonomously runs security tools on a Kali attack box, analyzes results, and iterates to find vulnerabilities.