Summary

pwno-mcp is a stateful system for autonomous pwn and binary research, engineered for LLM agents. It provides an isolated, controlled environment for debugging and exploit development, enabling efficient, automated binary analysis for security researchers.

How It Works

The core approach involves running GDB with pwndbg within an isolated Docker container, exposing stateful debugging, exploit I/O, and tooling via an MCP interface. This design allows LLM agents programmatic interaction, offering deterministic execution control, rapid context snapshots (registers, stack, disassembly, source, backtrace), and interactive exploit-driver workflows.

Quick Start & Requirements

Installation is Docker-based. A local workspace directory is needed for target binaries. The primary command runs a container with elevated privileges (SYS_PTRACE, SYS_ADMIN) and relaxed security (seccomp=unconfined, apparmor=unconfined), mounting the workspace.

docker run --rm -p 5500:5500 \
  --cap-add=SYS_PTRACE \
  --cap-add=SYS_ADMIN \
  --security-opt seccomp=unconfined \
  --security-opt apparmor=unconfined \
  -v "$PWD/workspace:/workspace" \
  ghcr.io/pwno-io/pwno-mcp:latest

The default MCP endpoint is http://127.0.0.1:5500/mcp. Comprehensive documentation is at docs.pwno.io.

Highlighted Details

• Stateful GDB + pwndbg debugger sessions.
• Deterministic execution control via GDB/MI.
• Fast context snapshots (registers, stack, disassembly, source, backtrace).
• Interactive exploit-driver workflows with pwncli.
• Multi-session support for parallel workflows.
• Workspace automation helpers (commands, processes, Python, RetDec).
• HTTP and stdio transport support.

Maintenance & Community

Issues and pull requests are welcomed. The README does not specify community channels (e.g., Discord, Slack) or details on core contributors or sponsorships.

Licensing & Compatibility

Licensed under CC BY-NC-ND 4.0 (Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International). This license prohibits commercial use and derivative works. Commercial licensing inquiries should be directed to oss@pwno.io.

Limitations & Caveats

The CC BY-NC-ND 4.0 license imposes significant restrictions, prohibiting commercial use and modifications, thus limiting adoption to non-commercial research and personal projects. The system is optimized for LLM agent integration, with advanced features like WebSocket streaming, memory analysis, heap exploitation, ROP generation, and symbolic execution listed as future enhancements.