plantuml-skill

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow ingests outsider-authored PlantUML text supplied by the user (Step 3 .puml generation from the user description) and then posts it to Kroki (curl -X POST https://kroki.io/plantuml/{png,svg}) for rendering, meaning free text reaches an LLM-untrusted rendering/execution context.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill includes an explicit "sudo apt install graphviz" instruction (SKILL.md:60), which tells the agent/user to perform a privileged system install and thus encourages modifying the machine state with elevated privileges.