og-image-design

Warn

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill guides users to install an external CLI tool via npx skills add belt-sh/cli and points to a remote installation guide at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. These resources originate from an external vendor not present in the verified trusted list.- [COMMAND_EXECUTION]: The skill leverages the Bash tool to execute belt commands. It provides templates for commands like belt login and belt app run that interpolate complex JSON payloads containing user-controlled HTML and prompt strings directly into shell executions.- [REMOTE_CODE_EXECUTION]: The skill's primary functionality relies on belt app run to execute remote application logic (e.g., infsh/html-to-image, falai/flux-dev-lora) on the inference.sh platform, delegating computation to third-party infrastructure.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of HTML and text prompts for image rendering. It lacks explicit boundary markers or instructions for the agent to sanitize this content before it is processed by remote tools, creating a surface for indirect injection attacks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 17, 2026, 04:09 PM
Security Audit — agent-trust-hub — og-image-design