awesome-content-graphics
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from various sources including URLs, text, and files. Maliciously crafted content could attempt to influence the agent's behavior.
- Ingestion points: SKILL.md Step 1 (URL fetching, file reading, image analysis).
- Boundary markers: The agent is instructed to state the subject in one sentence for user confirmation before starting the generation process.
- Capability inventory: Writing to temporary directories, fetching external URLs, and generating markup for local rasterization.
- Sanitization: The skill contains explicit 'Hard rules' (Rule 7) and instructions (Step 3) prohibiting the access of credentials, .env files, or environment variables.
- [EXTERNAL_DOWNLOADS]: Step 1 explicitly fetches data from user-supplied URLs, which can be used to pull content into the agent's context.
- [DYNAMIC_EXECUTION]: The agent creates markup (such as SVG or HTML) which is then rasterized to PNG. If the rasterization tool (e.g., a browser or image utility) handles this markup without sufficient isolation, it could be exploited to perform server-side request forgery (SSRF) or local file inclusion.
- [COMMAND_EXECUTION]: The skill relies on local system tools to rasterize markup into PNG images and to perform OCR to verify the text content of the generated images, necessitating the execution of subprocesses.
Audit Metadata