Security audit
Security checks for vulnerabilities and agentic risk
This Java checklist skill is not clearly malicious, but it requests read/exec/write authority and includes broad modify/delete/deployment/network instructions that are not well scoped to its stated code-review purpose.
Review this skill before installing. It appears to be a Java coding checklist, but it asks for broader read/exec/write capabilities and contains vague state-changing and network-troubleshooting instructions. Only use it where you are comfortable with an agent potentially running commands or writing files, and prefer a version that narrows activation to Java code review and removes delete/reset/deployment/network guidance.
The skill advertises create, modify, delete, save, export, import, and reset style operations through parameters and output handling, but provides no warnings or restrictions around file or system impact. In combination with declared read/exec/write tools, this creates a plausible path for destructive or unauthorized state changes that are unrelated to a passive Java rule-checking tool.
The skill states it is not suitable for operations or deployment management, yet later introduces command-line execution and network diagnostic guidance such as ping-based troubleshooting. This mismatch can mislead users and agents about the skill’s actual behavior boundary, increasing the chance that an agent invokes operational commands outside the intended Java self-check scope.
The trigger condition covers broad activities like code generation, programming assistance, debugging, testing, and deployment, which far exceed the stated purpose of checking 14 Java pitfalls. Overbroad activation makes it easier for the skill to be invoked in unrelated contexts where its write/exec capabilities could have unintended side effects.
For a skill advertised as a Java coding rules checklist, enabling exec-driven command execution and network diagnostics is unnecessary and expands the attack surface significantly. An attacker or prompt-injected user input could leverage this mismatch to trigger shell commands or external connectivity under the guise of code assistance.
