adithyan-ak/AgentHound

View on GitHub ↗

Offensive security framework for AI agent infrastructure - recon, credential looting, model exfiltration, poisoning, and attack-path analysis across MCP, A2A, gateways, and AI services. BloodHound for the agentic stack.

264 ★57 forksGoUpdated 23d ago

What you need to know

A browser-based assistant MCP AI with Vision and V8 APIs for searching, reading and exploring the web through a real browser port.

Install

Configure the AgentHound MCP server in your client with browser access enabled

Usage

  • Use the browser API to search and read pages, and the Vision API for visual understanding of pages

Key features

  • Browser-based web searching and reading
  • Vision API for visual page understanding
  • V8 API for in-browser scripting

Best for

Users who want a browsing-aware assistant that can search, read and visually inspect web pages through MCP.

Platforms: Web · BrowserClients: MCP clients

Reviewed 2026-08-11

Topics

a2aagentic-aiai-agentsai-securityattack-pathsbloodhoundllm-securitymcpmodel-context-protocolneo4joffensive-securitypentestingred-teamsecurity-tools
Stars
264★
Forks
57
Language
Go
License
Apache-2.0
Created
2026-04-05
Last push
2026-08-12