qzcli

Warn

Audited by Gen Agent Trust Hub on Nov 10, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's installation instructions require cloning a repository from an unverified third-party GitHub account (https://github.com/tianyilt/qzcli_tool) that does not belong to a recognized or trusted organization.\n- [REMOTE_CODE_EXECUTION]: The installation process executes code from the cloned repository using 'pip install -e .', allowing for arbitrary code execution within the environment during the installation phase.\n- [CREDENTIALS_UNSAFE]: Documentation provides examples of passing sensitive passwords directly as command-line arguments (e.g., qzcli login -u USER -p 'PASSWORD'). This practice risks exposing credentials in cleartext within shell history files like ~/.bash_history.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes a 'batch' command that processes external JSON configuration files to automate job submissions. These files can contain command templates that are executed via shell, creating a surface for indirect prompt injection if an attacker provides a malicious configuration file.\n
  • Ingestion points: Processes 'batch_config.json' file content through the 'qzcli batch' command.\n
  • Boundary markers: No boundary markers or instructions to ignore embedded commands are present in the documentation.\n
  • Capability inventory: Access to 'Bash(*)', 'Read', and 'Write' tools allows the execution of arbitrary shell commands and file modifications.\n
  • Sanitization: No evidence of input validation or sanitization for the command templates defined in configuration files.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 10, 2026, 12:12 PM