search

Overview

LogoGitHub - six2dez/burp-ai-agent: Burp Suite extension that adds built-in MCP tooling, AI-assisted analysis, privacy controls, passive and active scanning and moreGitHubchevron-right

The bridge between Burp Suite and modern AI.

Burp AI Agent is an extension for Burp Suite that integrates AI capabilities into your security workflow. It offers:

  • Pluggable Backends: Use local models (Ollama, LM Studio), generic OpenAI-compatible providers, or cloud providers (Gemini, Claude, OpenAI/Codex, OpenCode). Add custom backends via drop-in JARs.

  • Privacy-First Design: Configurable redaction modes (Strict/Balanced/Off) to scrub sensitive data before it leaves Burp.

  • MCP Server: A full Model Context Protocol (MCP) server with 53+ tools that allows external AI agents (like Claude Desktop) to interact with Burp — browsing history, sending requests, scanning, and creating issues autonomously.

  • AI Scanners: Passive and Active scanners that autonomously analyze traffic and flag vulnerabilities across 62 vulnerability classes.

  • Audit Logging: JSONL-based logging with SHA-256 integrity hashing for compliance and reproducibility.

hashtag
Key Features

Feature
Description

7 Built-in Backends

Ollama, LM Studio, Generic OpenAI-compatible, Gemini CLI, Claude CLI, Codex CLI, OpenCode CLI.

53+ MCP Tools

History, Repeater, Intruder, Scanner, Scope, Site Map, Collaborator, Utilities, and more.

62 Vulnerability Classes

From SQLi and XSS to cache poisoning, JWT attacks, and API security issues.

3 Scan Modes

BUG_BOUNTY, PENTEST, and FULL for different engagement types.

3 Privacy Modes

STRICT (zero trust), BALANCED (pragmatic), OFF (raw data).

9 Prompt Templates

Customizable templates for all context menu actions.

Burp Pro Integration

Native ScanCheck, Collaborator OAST, scanner issue actions.

hashtag
Why use this?

  1. AI-Assisted Analysis: Use AI to analyze requests, explain JS, or draft proof-of-concept (PoC) exploits.

  2. Local Privacy: Run Llama 3 or Mistral locally via Ollama for zero-data-leakage analysis.

  3. Agentic Workflows: Connect Claude Desktop to Burp via MCP to let the AI "drive" Burp — navigating the site map, sending test requests, and verifying issues autonomously.

  4. Automated Scanning: Let passive and active AI scanners work in the background while you focus on manual testing.

  5. Compliance Ready: Audit logging with SHA-256 hashing, deterministic redaction, and configurable privacy modes.

hashtag
Getting Started

hashtag
Documentation

  • User Guide: Tour the UI, context menus, chat, and prompt templates.

  • Scanners: Passive and Active AI scanning.

  • MCP Reference: Learn how to connect external agents.

  • Privacy: Understand how your data is protected.

  • Examples: Real-world workflows and sample prompts.

  • Reference: Complete settings, glossary, and troubleshooting.

  • Developer: Architecture, data flow, and extension development.

hashtag
Operational Notes

  • Passive scanner reliability hardening (atomic state, bounded backend startup polling, latch-based completion wait).

  • Active scanner queue backpressure with a hard cap (default: 2000 queued targets) to prevent unbounded growth.

  • Shared HTTP backend support for retry/client/history logic across Ollama, LM Studio, and OpenAI-compatible providers.

  • Type-safe scanner settings in code (SeverityLevel, PayloadRisk, ScanMode) with backward-compatible persistence.

  • AGENTS profile validation against enabled MCP tools, with warnings in Settings.

  • Settings UI modularization into dedicated panel classes (Backend, Passive, Active, MCP, Prompt, Privacy, Help).

  • Passive scanner issue details now include model reasoning when provided by backend JSON output.

  • Structured prompt defaults moved to role/task/scope/output format for better response consistency.

  • Additional unit tests for scanner parsing/generation/analyzers and backend conversation history.

NextInstallation

Last updated