interrogate

Pass

Audited by Gen Agent Trust Hub on Nov 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to spawn sub-agents that "run commands in its own worktree" and later "rerun" those commands for verification. This capability allows the AI to execute arbitrary shell commands it generates based on its analysis of the code.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including PR descriptions and commit messages. Since the agent executes commands based on its interpretation of this data, it is potentially vulnerable to malicious instructions embedded within code diffs or PR metadata.\n
  • Ingestion points: PR descriptions, commit messages, and repository diffs referenced in SKILL.md.\n
  • Boundary markers: The instructions provide a specific task brief but lack delimiters or security instructions to prevent the agent from being influenced by instructions hidden in the data it processes.\n
  • Capability inventory: The agent is authorized to run shell commands and generate/execute repro scripts as defined in SKILL.md.\n
  • Sanitization: No sanitization or validation of the generated shell commands is performed prior to execution.\n- [DYNAMIC_EXECUTION]: The skill workflow involves creating and running "repro script[s]" and "prototype[s]" at runtime. These scripts are generated based on the untrusted input data under review, creating a vector for dynamic code execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 17, 2026, 09:14 AM