skills/shihyuho/skills/sdkman/Gen Agent Trust Hub

sdkman

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill's core functionality involves executing shell commands to manage environment variables and run build tools (e.g., sdk use, mvn, gradle). These operations are restricted to the current shell session to prevent state leakage.
  • [EXTERNAL_DOWNLOADS]: The skill can trigger downloads and installations of SDK candidates via SDKMAN. The instructions explicitly require the agent to obtain user approval before performing any persistent downloads or installations.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests configuration data from project-local files such as .sdkmanrc and .java-version.
  • Ingestion points: Configuration files are read from the workload directory or its parent directories within the same Git worktree (specifically SKILL.md in the 'Resolve the workload context' section).
  • Boundary markers: The skill defines the worktree root as the environment boundary and requires validation of the file path before use.
  • Capability inventory: The skill uses sdk env and shell commands to apply the discovered configurations, as detailed in the 'Run with an ephemeral version' section of SKILL.md.
  • Sanitization: The agent is instructed to follow exact declarations from the files and to verify the presence of candidates before execution, stopping for user confirmation if requirements are not met.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:59 PM