tool-advisor
Fail
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill's primary discovery phase (Phase 1) requires the agent to execute a complex Bash script that enumerates the user's local filesystem to identify MCP servers, skills, agents, and installed dev tools.
- [CREDENTIALS_UNSAFE]: The discovery script specifically targets and reads sensitive configuration files that often contain API keys or secrets, including
~/.claude/settings.json,~/.codex/config.json, and.envfiles. It reports the presence and line count of these environment files directly into the agent's context. - [DYNAMIC_EXECUTION]: The shell script utilizes several dynamically generated Python snippets executed via
python3 -cto parse JSON and TOML configuration data extracted from the local system.
Recommendations
- AI detected serious security threats
Audit Metadata