msoedov/agentic_security
An open-source vulnerability scanner for agentic AI systems that tests LLMs and agent workflows against jailbreaks, fuzzing, and multimodal attacks.
Not currently ranked — collecting fresh signals.
star history
Agentic Security provides a red teaming toolkit for evaluating the security of AI agents and LLM-based systems. It simulates multi-step jailbreak attacks, probes vulnerabilities across text, image, and audio inputs, and conducts comprehensive fuzz testing to identify weaknesses in safety mechanisms. The toolkit integrates with various LLM APIs for stress testing and serves as a security evaluation framework for developers building agentic applications.
Frequently asked
- What is msoedov/agentic_security?
- An open-source vulnerability scanner for agentic AI systems that tests LLMs and agent workflows against jailbreaks, fuzzing, and multimodal attacks.
- Is agentic_security open source?
- Yes — msoedov/agentic_security is open source, released under the Apache-2.0 license.
- What language is agentic_security written in?
- msoedov/agentic_security is primarily written in Python.
- How popular is agentic_security?
- msoedov/agentic_security has 1.9k stars on GitHub.
- Where can I find agentic_security?
- msoedov/agentic_security is on GitHub at https://github.com/msoedov/agentic_security.