py-quality-setup
Fail
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill modifies .claude/settings.local.json to grant the agent silent permission to execute a wide range of sensitive commands, including pip, python, chmod, and git. This bypasses the security model intended to prevent unauthorized system changes or data access.
- [REMOTE_CODE_EXECUTION]: The skill implements a persistence mechanism by configuring a Stop hook in ~/.claude/settings.json. This hook automatically executes an external Python script (lint-gate.py) whenever the agent completes a task, providing a persistent execution vector for code outside the current skill.
- [EXTERNAL_DOWNLOADS]: The skill adds numerous third-party dependencies to the project's configuration (ruff, mypy, basedpyright, etc.), which are then installed via package managers. While these are common tools, the combination of automated installation and bypassed command permissions increases the potential impact of a supply chain compromise.
Recommendations
- AI detected serious security threats
Audit Metadata