rock-debug

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the rockcli command-line utility to perform diagnostic and troubleshooting actions.
  • The exec subcommand is used to run shell commands inside the sandbox environment, such as ps aux, netstat, and resource checks (df -h, free -h).
  • The replay subcommand facilitates the execution of historical command sequences in a new sandbox to replicate and verify reported issues.
  • [PROMPT_INJECTION]: The skill relies on the agent analyzing sandbox logs and command history, which presents a surface for potential indirect prompt injection.
  • Ingestion points: The agent reads untrusted data from log files (command.log, access.log, rocklet.log) and the output of the history command as described in SKILL.md.
  • Boundary markers: The instructions do not provide specific delimiters or ignore-instructions to help the agent distinguish between legitimate log data and potentially malicious embedded instructions.
  • Capability inventory: The skill grants the agent the ability to execute commands through the rockcli exec and rockcli replay tools.
  • Sanitization: There is no evidence of log content sanitization or validation before the agent processes the data for debugging analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 07:54 AM
Security Audit — agent-trust-hub — rock-debug