codex-theme-maker
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several system commands to manage the theme development lifecycle, including
git clonefor repository setup, Node.js scripts for auditing and packaging, and Python scripts for image processing tasks such as alpha normalization and contact sheet generation. - [EXTERNAL_DOWNLOADS]: Fetches components and configuration from the author's public GitHub repository (
github.com/Wangnov/awesome-codex-skins) and leverages an external image generation utility (gpt-image-2-skill). - [DYNAMIC_EXECUTION]: Implements a workflow for real-time application skinning by dynamically assembling and injecting CSS/HTML payloads into the Codex desktop application via the Chrome DevTools Protocol (CDP).
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect injection as it processes untrusted user-provided concept art and IP style descriptions to influence the theme generation process. • Ingestion points: Visual concept art and IP descriptions provided by the user in Phase 0 and 1. • Boundary markers: The instructions do not explicitly define boundary markers for the user-provided prompts, relying instead on post-generation validation scripts. • Capability inventory: Includes subprocess execution (Node.js, Python, Git), filesystem operations (theme and config file writing), and application interaction via CDP. • Sanitization: Employs dedicated scripts (
audit-theme.mjs,verify-alpha.py) to validate theme schemas, ensure file path safety, and enforce asset integrity standards.
Audit Metadata