ai-fuzzing
Audited by Socket on Sep 19, 2026
3 alerts found:
Securityx2AnomalySUSPICIOUS. The skill is internally consistent for fuzzing, with proportionate tool use and no evident credential harvesting or third-party exfiltration. However, it is a high-risk offensive-security skill for AI agents and includes explicit defense-evasion guidance, which substantially elevates security risk despite benign-looking data flows.
The fragment is an offensive-security and robustness-fuzzing guide with executable examples. It contains substantial operational and availability risk when directed at unauthorized or production systems, especially the flooding, slowloris, oversized-payload, and arbitrary-ioctl tests. It does not show clear malware, credential theft, persistence, exfiltration, or obfuscated code. Use only with explicit authorization and isolated targets.
The fragment documents and automates potentially intrusive API and authentication testing, including JWT bypass attempts, credential brute forcing, rate-limit evasion tests, and state-changing CSRF probes. This creates operational and authorization risk if used against systems without permission, but there is no clear evidence of malware, covert exfiltration, persistence, or supply-chain sabotage in the visible code.