creative-director
Warn
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions facilitate the download and installation of a large external repository from 'https://github.com/smixs/creative-director-skill'. This source is not a recognized trusted vendor, and the installation includes over 600 files.\n- [REMOTE_CODE_EXECUTION]: The installation process uses 'npx skills add' to fetch and execute code from an external GitHub repository, which allows for remote code to be added to the agent's environment.\n- [COMMAND_EXECUTION]: The SKILL.md file contains bash commands designed to be executed to check for existing installations and to perform the installation using the 'npx skills add' command.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing external data such as text, notes, and PDFs. This creates a surface for indirect prompt injection where malicious instructions embedded in a brief could influence the agent's behavior during the ideation or scoring phases.\n
- Ingestion points: Processes text, notes, and PDFs supplied by the user (SKILL.md).\n
- Boundary markers: None present in the provided instructions to distinguish untrusted content.\n
- Capability inventory: Executes shell commands via 'npx' as part of its setup and operation (SKILL.md).\n
- Sanitization: No evidence of input validation or sanitization for processed briefs.
Audit Metadata