---
title: "RAPTOR Turns Claude Into an Autonomous Hacking Agent With 3,000 Stars"
subtitle: "RAPTOR turns Claude Code into a full autonomous security agent that scans, validates, exploits, and patches vulnerabilities — now with multi-model support and Z3 constraint solving"
url: "https://alphasignal.ai/news/raptor-turns-claude-into-an-autonomous-hacking-agent-with-3-000-stars"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-06-23T11:58:03.000Z"
updated: "2026-06-25T13:09:04.592Z"
type: "Repo"
topics:
  - "Agents"
  - "Security"
subtopics:
  - "Computer Use"
  - "Tool Use"
  - "Red Teaming"
upvotes: 102
source_url: "https://github.com/gadievron/raptor"
comments: 0
truncated: true
---

# RAPTOR Turns Claude Into an Autonomous Hacking Agent With 3,000 Stars

> RAPTOR turns Claude Code into a full autonomous security agent that scans, validates, exploits, and patches vulnerabilities — now with multi-model support and Z3 constraint solving

[RAPTOR](https://github.com/gadievron/raptor) (Recursive Autonomous Penetration Testing and Observation Robot) is an open-source framework that turns Claude Code into a fully autonomous security research agent. Built on top of Claude Code, it chains together static analysis, binary analysis, LLM-powered vulnerability validation, exploit generation, and patch writing into a single workflow you can run against a codebase or binary. The project has crossed 3,000 GitHub stars and is actively maintained, with a v3.0.0 release that adds multi-model orchestration, Z3 constraint solving, and a software composition analysis pipeline.

## A problem the defenders didn't have an answer to

The researchers behind RAPTOR , Gadi Evron (CEO of Knostic), Daniel Cuthbert (Black Hat Review Board member), Thomas Dullien (known as Halvar Flake), and Michael Bargury (CTO of Zenity) , built it after seeing how powerful coding agents could be and wanting to harness that capability for security research. Their motivation was blunt: "The bottom line is, attackers are using AI to find vulnerabilities, exploit vulnerabilities, and orchestrate and automate their attacks. On the defensive side, we don't have that yet."

The timing is notable. Cloudflare recently published a detailed breakdown of their own internal AI vulnerability discovery harness, and the RAPTOR team was quick to point out the convergence. Cloudflare detailed a model-agnostic AI security harness architecture for scalable vulnerability discovery and validation, emphasizing an architecture where models are treated as interchangeable components , moving beyond relying on single, frontier AI models. RAPTOR had been doing exactly this, independently, for months.

## Two layers that do very different things

The architecture is the key insight. RAPTOR splits responsibility cleanly between a deterministic execution layer and an LLM reasoning layer , and that split is what makes it composable.

* The **Python execution layer** (`raptor.py`, `packages/`, `core/`, `engine/`) handles the heavy lifting: running Semgrep and CodeQL, managing subprocesses, parsing SARIF, deduplicating findings, dispatching LLM API calls, tracking costs, writing output files.
* The **Claude Code decision layer** (`.claude/`, `tiers/`, `CLAUDE.md`) makes the calls: which findings to prioritize, how to interpret results, what the attack scenario is, whether the exploit is realistic , implemented as Claude Code skills, commands, and agents that load progressively.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [gadievron/raptor — GitHub](https://github.com/gadievron/raptor)
- [New Raptor Framework Uses Agentic AI to Create Patches — Dark Reading](https://www.darkreading.com/vulnerabilities-threats/new-raptor-framework-uses-agentic-ai-create-patches)
- [Cloudflare's AI Security Blueprint — StartupHub.ai](https://www.startuphub.ai/ai-news/technology/2026/cloudflare-s-ai-security-blueprint)
- [AI Pentesting Agents 2026: The Rise of 39+ Tools Tested — AppSec Santa](https://appsecsanta.com/research/ai-pentesting-agents-2026)
- [\[tl;dr sec\] #307 — AI Bug Hunting Tools](https://tldrsec.com/p/tldr-sec-307)