RAPTOR is an open-source security research framework designed for both offensive and defensive purposes. Built on Claude Code, it automates various security tasks to streamline code analysis and vulnerability management. Key features include scanning code with tools like Semgrep and CodeQL, fuzzing binaries with American Fuzzy Lop, and utilizing advanced LLM reasoning to analyze vulnerabilities and generate exploit proofs. It also offers patching capabilities, including specific fixes for recent vulnerabilities in software like FFmpeg. The framework aims to enhance security operations with a focus on automation and community contributions. Recent updates have added significant capabilities, such as a unified LLM interface for streamlined communication and cost tracking, as well as improved budget management features. RAPTOR’s GitHub forensics functionalities allow users to collect evidence from multiple sources, recover deleted content, and generate detailed forensic reports. The architecture supports multi-agent orchestration, enabling parallel evidence gathering and sequential analysis, which can be particularly useful for investigating security incidents. Installation options include setting up a local environment or using a pre-configured devcontainer. The devcontainer includes all necessary tools and dependencies, though it is sizable at around 6GB. Users can initiate interactions with RAPTOR simply by saying "hi" and can start testing with sample data provided in the repository. The framework emphasizes safety, requiring user confirmation for potentially dangerous operations while auto-executing safer tasks. Overall, RAPTOR combines traditional security methodologies with modern automation to enhance the efficiency and effectiveness of security research.