What is awesome-offensive-ai-agentic-landscape?

This repository is a curated reference list covering the intersection of artificial intelligence and offensive cybersecurity. It compiles open-source projects, academic papers, evaluation benchmarks, and commercial products in four main areas: AI-assisted penetration testing (the practice of probing computer systems for security weaknesses), red teaming of language models (testing AI systems for vulnerabilities and unsafe behavior), autonomous agents that can carry out attacks, and automated vulnerability discovery. The list was assembled with a data cutoff of May 2026 and is documented in both Chinese and English. It covers 61 open-source projects, 67 academic papers from June 2023 through January 2026, 12 capability benchmarks, 6 other curated reference lists, and 48 commercial products including 39 international and 9 Chinese vendors. Entries with at least 1,000 stars on GitHub have their star count displayed in thousands. The project tables are sorted by GitHub star count. The most-starred open-source projects include tools for automated web application security testing, LLM vulnerability scanning (described in the README as doing for language models what the network scanner nmap does for networks), fully autonomous penetration testing agents, and frameworks for testing whether AI systems can be tricked through crafted inputs. Several entries include projects from well-known institutions and companies such as NVIDIA, Microsoft, OWASP, and Trail of Bits. The intended audience is security researchers, security engineers, and enterprise decision-makers who want an overview of where AI capabilities are being applied in offensive security. The document is maintained as a living reference and accepts community contributions to fill in missing entries across each category. The full README is longer than what was shown.