oz-finish-task

Pass

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions direct the agent to run a shell command using a path provided by the $OZ_CLI environment variable. This facilitates reporting task results to the Oz platform.
  • [INDIRECT_PROMPT_INJECTION]: The shell command template includes a <summary> field that processes task descriptions, creating a potential surface for indirect injection.
  • Ingestion points: The <summary> placeholder in the shell command found in SKILL.md.
  • Boundary markers: The summary value is wrapped in single quotes.
  • Capability inventory: Execution of shell commands via the Oz CLI tool.
  • Sanitization: The instructions do not specify sanitization or escaping techniques for the summary text, which could lead to shell command breakage if single quotes are present in the input.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 23, 2026, 04:59 AM
Security Audit — agent-trust-hub — oz-finish-task