Skills
skills/luongnv89/skills/install-script-generator/Gen Agent Trust Hub

install-script-generator

Fail

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill generates and promotes installation commands that pipe remote script content from GitHub raw URLs directly into a shell (e.g., curl | bash). This pattern is a significant security risk as it executes unverified remote code.
  • [COMMAND_EXECUTION]: The scripts/executor.py script uses subprocess.run(shell=True) to execute arbitrary commands from a YAML plan. The skill also performs automated git fetch and git pull operations, which execute code from a remote repository.
  • [EXTERNAL_DOWNLOADS]: The skill downloads and executes scripts from raw.githubusercontent.com. While the platform is well-known, the specific content is user-controlled and unverified.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. Ingestion points: Repository files like Makefile, package.json, and setup.py are analyzed in SKILL.md. Boundary markers: Absent. Capability inventory: executor.py can run arbitrary shell commands with shell=True. Sanitization: Absent. Malicious project files could inject instructions into the generated installation plan.
  • [DATA_EXFILTRATION]: scripts/env_explorer.py collects detailed system metadata (OS, architecture, permissions, paths). This information is stored locally in env_info.json but could be leaked if included in documentation or scripts generated by the agent.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/ - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 11, 2026, 10:46 PM