configure-ecc
Fail
Audited by Gen Agent Trust Hub on Apr 14, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill downloads content from an unverified GitHub repository and installs it into persistent skill directories (~/.claude/skills/). Files in these directories are automatically loaded and executed by the agent platform, resulting in remote code execution from an unverified source.\n- [EXTERNAL_DOWNLOADS]: Fetches code from a third-party GitHub repository (affaan-m/everything-claude-code) that is not a recognized or trusted source, introducing potential supply chain vulnerabilities.\n- [COMMAND_EXECUTION]: Utilizes multiple shell commands including
git clone,rm,mkdir, andcpto manage file system operations and repository cloning. These commands provide a broad surface for manipulating local files.\n- [PROMPT_INJECTION]: The optimization process in Step 5 creates a surface for indirect prompt injection by reading and editing remote instructions without sanitization or boundary markers. 1. Ingestion Point: Remote SKILL.md files (SKILL.md). 2. Boundary Markers: Absent. 3. Capability Inventory: File-read and file-write operations across the installation target (SKILL.md). 4. Sanitization: Absent.
Recommendations
- AI detected serious security threats
Audit Metadata