codex-image-gen
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands via
codex execto invoke image generation tools with specific sandbox and network permissions. - [INDIRECT_PROMPT_INJECTION]: The skill constructs complex English prompts based on user queries (
{ENGLISH_PROMPT}). While used for image generation, this surface could be exploited to manipulate the agent's logic if instructions are embedded in the query, though the risk is localized to the tool output. - [EXTERNAL_DOWNLOADS]: Mentions uploading data to Cloudflare R2 and GitHub raw URLs using
wranglerandghCLI tools; these are classified as well-known/trusted services and do not escalate the verdict. - [SAFE]: The skill uses placeholders for paths and configurations, and its instructions for local authentication file checks are standard troubleshooting steps.
Audit Metadata