cso
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to run shell commands for stack detection, secret scanning (git log), and infrastructure auditing.
- [EXTERNAL_DOWNLOADS]: It depends on external binaries located at specific paths in the user's home directory (~/.claude/skills/gstack/bin/), including 'gstack-config', 'gstack-learnings-search', and 'gstack-learnings-log'. These are non-standard utilities used for cross-project intelligence.
- [DATA_EXFILTRATION]: To perform its audit, the skill intentionally reads sensitive project data such as .env files, git logs (searching for keys like 'AKIA', 'sk-', 'ghp_'), and CI/CD workflow configurations.
- [PROMPT_INJECTION]: The skill includes a 'Anti-manipulation' rule instructing the AI to ignore any instructions found within the audited codebase, which serves as a security measure against indirect prompt injection.
- [DATA_EXFILTRATION]: The tool can scan globally installed agent skills and hooks outside the current repository if the user grants permission, potentially accessing sensitive configuration data across the host system.
Audit Metadata